Full text of "Motivation of behavior"

See other formats

UNIVERSITY
OF FLORIDA
LIBRARIES

Digitized by the Internet Archive

in 2010 with funding from

Lyrasis IVIembers and Sloan Foundation

http://www.archive.org/details/motivationofbehaOObrow

The Motivation of Behavior

McGRAW-HILL SERIES IN PSYCHOLOGY
Harry F. Harlow, Consulting Editor

Barker, Kounin, and Wright • Child Behavior and Development
Beach, Hebb, Morgan, and Nissen • The Neuropsychology of

Lashley
von Bekesy • Experiments in Hearing
Berlyne • Conflict, Arousal, and Curiosity
Blum • Psychoanalytic Theories of Personality
Brov^^n • The Motivation of Behavior
Brown • The Psychodynamics of Abnormal Behavior
Brown and Ghiselli • Scientific Method in Psychology
Cattell • Personality
Crafts, Schneirla, Robinson, and Gilbert • Recent Experiments

in Psychology
Deese • The Psychology of Learning
DoLLARD AND MiLLER • Personality and Psychotherapy
DoRcus AND Jones • Handbook of Employee Selection
Ferguson • Personality Measurement

Ferguson • Statistical Analysis in Psychology and Education
Ghiselli and Brown • Personnel and Industrial Psychology
Gilmer • Industrial Psychology
Gray • Psychology Applied to Human Affairs
Gray • Psychology in Industry

Guilford • Fundamental Statistics in Psychology and Education
Guilford • Personality
Guilford • Psychometric Methods
Haire • Psychology in Management
Hirsh • The Measurement of Hearing
HuRLOCK • Adolescent Development
HuRLOCK • Child Development

HuRLOCK • Developmental Psychology

Johnson • Essentials of Psychology

Karn and Gilmer • Readings in Industrial and Business Psychology

Krech and Crutchfield • Theory and Problems of Social Psychology

Lewin • A Dynamic Theory of Personality

Lewin • Principles of Topological Psychology

Lewis • Quantitative Methods in Psychology

Maier and Schneirla • Principles of Animal Psychology

Miller • Language and Communication

MisiAK AND Staudt • Catholics in Psychology: A Historical Survey

Moore • Psychology for Business and Industry

Morgan and Stellar • Physiological Psychology

Page • Abnormal Psychology

Reymert • Feelings and Emotions

Seashore • Psychology of Music

Shaffer and Lazarus • Fundamental Concepts in Clinical Psy-
chology

SiEGEL • Nonparametric Statistics: For the Behavioral Sciences

Stagner • Psychology of Personality

Tovv^nsend • Introduction to Experimental Method

Vinacke • The Psychology of Thinking

Wallen • Clinical Psychology: The Study of Persons

Waters, Rethlingshafer, and Caldwell • Principles of Com-
parative Psychology

ZuBEK and Solberg * Humau Development

John F. Dashiell was Consulting Editor of this series from its in-
ception in 1931 until January I, 1950. Clifford T. Morgan was Con-
sulting Editor of this series from January 1, 1950 until January 1, 1959.

THE MOTIVATION
OF BEHAVIOR

(co- m bl Z.

JUDSON SEISE BROWN

Graduate Research Professor
University of Florida

McGRAW-HILL BOOK COMPANY, INC,

New York Toronto London 1961

THE MOTIVATION OF BEHAVIOR

Copyright © 1961 by the McGraw-Hill Book Company, Inc. Printed
in the United States of America. All rights reserved. This book, or
parts thereof, may not be reproduced in any form without permission
of the publishers. Library of Congress Catalog Card Number 60-14612

08305

PREFACE

Two major aims have served as guideposts throughout the
writing of this book. The more important of these has been to
develop a tightly reasoned, systematic analysis of the concept of
motivation, with special emphasis upon its relative utility as an
explanatory component of general behavior theory. The lesser
goal has been to formulate the analysis in such a manner as to
make it intelligible to the advanced undergraduate or beginning
graduate student in psychology.

In principle, if all behavior is motivated, any book about mo-
tivation should encompass all of psychology, at the very least.
Since such inclusiveness is patently impossible, an author is forced,
in reducing his task to manageable proportions, to neglect certain
topics of special interest to particular readers. He must be selec-
tive, therefore, and in the absence of widely accepted criteria for
evaluating the importance of each and every area, the process
of selection must be governed primarily by idiosyncratic factors.
In the succeeding pages, for example, almost no consideration is
given to the topic of sexual behavior. This omission should not
be construed, however, as reflecting the belief that there are no
significant motivational aspects to sexual behavior. Rather it is a
consequence of the conviction that while much is known of the
effects of hereditary and environmental variables upon sexual be-
havior, especially in lower mammals, the concept of sexual motiva-
tion has thus far played only the vaguest of roles in behavior
theory generally. This point is underscored by the observation
that the terms drive and motivation are almost never used in
published works on reproductive activities, even when such works

viii PREFACE

have been written by psychologists. To be sure, students of repro-
ductive activities are concerned with motivated behavior, but they
have shown httle interest in the construct of sexual motivation.
Similarly detailed justifications to support my neglect of other
topics, such as hoarding behavior, taste preferences, and the vast
Freudian literature on unconscious motivation, would lengthen
this preface unduly. It need only be said, therefore, that in some
instances I was unable clearly to discern the motivational impli-
cations of the research, in certain cases my fund of knowledge
was inadequate, and in others the existing literature seemed both
adequate and extensive.

Some readers may feel that I have paid undue attention to
Hull's multiplicative-drive theory of motivation. Here I must
plead guilty to the belief that this conception, augmented by
other theorists' emendations, is the most explicitly structured of
existing views, and, at present, probably the most useful theory
of motivation we possess. So far as I am aware, no other con-
ception of the motivational process has generated so much ex-
perimental work or led to such varied theoretical inquiry. Never-
theless, on a number of occasions I have suggested, and some-
times defended, the possibility that Hull's conception is either
inadequate or noneconomical. Rather generally throughout the
book, and particularly in Chapter 4, serious and sympathetic at-
tention is given to alternative, nonmotivational interpretations of
the effects of so-called motivational variables upon behavior.

Grateful and enthusiastic acknowledgment is made to the many
individuals who, as the result of their careful and intelligent read-
ing of early versions of the manuscript, did much to improve its
clarity and precision. In this respect, I. E. Farber's penetrating
and constructive review of the entire manuscript was of inesti-
mable value. The unusually discerning and appropriate criticisms
of Bettina Bass, Richard deCharms, Kenneth P. Goodrich,
Leonard E. Ross, Charles C. Spiker, and Allan M. Wagner evoke
not only my warmest thanks but also, frankly, my envy. Robert
S. Witte, who used a preliminary draft of the manuscript in his
motivation classes at Cornell College, Mount Vernon, Iowa, re-
lieved some of my qualms concerning the suitability of the ma-
terial for undergraduates, and in addition made numerous positive

PREFACE ix

suggestions for the improvement of the manuscript. Kenneth W.
Spence repeatedly guided me to important references and concepts,
and our informal discussions have been of great value in helping
me toward clearer conceptions of difficult problems.

I am especially grateful to my wife, Julia S. Brown, for her
cheerful and painstaking assistance in the preparation of the final
manuscript and for her scholarly criticisms both of my phraseology
and of certain of my arguments. The high-level secretarial assist-
ance provided at various times by Betty Jean Stoner, Jean Hansen,
and Sandra Wood is a pleasure to acknowledge, as is the biblio-
graphic work of Marigold Belloni.

Some of that priceless commodity, time to think and to write,
was provided by the Graduate College of the State University of
Iowa in the form of a Graduate Research Professorship during
the period from September 1956 to January 1957. Much of the
theoretical development and some of the unpublished experi-
ments cited in Chapter 8 were supported by a small grant
(M-1789) from the National Institutes of Health.

Permission to reproduce figures and to make quotations were
granted by individual authors and their work is credited in each
instance as a citation to the list of references. Permission to quote
and reproduce figures was also granted by the University of
Nebraska Press, Yale University Press, the New York Academy of
Sciences, the Journal Press, the American Journal of Psychology,
the Journal of Personality, and the American Psychological Asso-
ciation. The cooperation of all of these authors and organizations
is gratefully acknowledged.

Judson S. Brown

CONTENTS

Preface vii

1 . The Psychologist's Task and the Problem of Motivation . 1

2. Intervening Variables and the Definition and Measurement

of Drive 28

3. Primary Sources of Drive 57

4. Motivational and Associative Interpretations of "Motivated"
Behavior 97

5. Learned Responses as Sources of Drive 138

6. Motivational Consequences of Frustration and Conflict . 194

7. Motivational Variables and Human Performance . 224

8. Motivational Variables and Perception 266

9. Miscellaneous Motivational Problems 327

References 365

Name Index 389

Subject Index 394

CHAPTER

miiiiiiiiiiiiii

The Psychologist's Task and
the Probfem of Motivation

Most present-day psychologists would probably agree that their
principal task is to study and to strive to understand the behavior
of living organisms, both human and subhuman. To accomplish
this aim, the psychologist, if he wishes to proceed in a scientific
manner, must begin by making dispassionate, careful, and re-
peated observations of the activities of organisms in a wide variety
of circumstances. He must observe the movements they make, the
sounds they utter, the objects they manipulate, and their interac-
tions with other organisms. Moreover, these observations must be
recorded in a relatively permanent manner so that they can be
evaluated and analyzed long after the behavior has ceased and the
process of observing has been terminated. Initially, therefore, the
psychologist's task is to observe the actions of organisms and to
record what he has seen in as precise and accurate a manner as
possible. These recorded bits of information constitute the basic
data with which he begins his work.

The psychologist concentrates his attention primarily upon the
behavior of living organisms, not upon their physiological func-

2 THE MOTIVATION OF BEHAVIOR

tions, their anatomical structure, or their physical-chemical com-
position. Psychological data are thus derived from one class of
events and the data of other sciences from different classes.
Because of this one may maintain, with some justifica-
tion, that the subject matter of psychology is unique in comparison
with the subject matter of other disciplines.

But we must also remember that there is almost nothing unique
or characteristically "psychological" about the methods by which
the psychologist collects his raw data. His observational techniques
are, in essence, the same as those of the zoologist, the physiologist,
and the physicist. Sometimes the only equipment he uses is a
pencil and paper. He notes whether a rat turns right or left in a
maze; he counts the motorists who do or do not stop for a red
traffic light; he tabulates the swallows a baby makes while drink-
ing a bottle of milk; he records the number of mistakes made by
a student on an intelligence test. On other occasions, however,
the psychologist's observations cannot be made without the help
of much more elaborate equipment. He may use a stop watch to
measure the time taken by a student in completing a question-
naire; he may employ an electronic instrument to detect and record
changes in the resistance of the skin due to loud sounds or painful
stimuli; he may use the physicist's gauges to measure the force
of a manual movement. But whether the equipment is simple
or complex, in all of these instances the psychologist is using the
well-known, well-standardized methods of the natural sciences.
He measures the vigor of a movement and expresses its force in
physical units such as grams or pounds; he times the duration
of a movement in seconds or minutes; he counts the frequencies
of events of many kinds and records the results in standard numeri-
cal notations. His raw data, therefore, when tabulated and recorded
are simply the frequencies with which certain events occur, their
magnitudes, their rates, or their durations. Considered in this
restricted sense, as frequency counts, quantities that change with
time, amplitude measurements, and the like, the basic data of
psychology are identical with the basic data of all other sciences.
Psychological data could not be obtained, of course, save for the
existence of living organisms, but the fundamental observations
are made by methods common to all sciences and are couched in

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 3

the language of the natural scientist, not in any special-purpose
psychological jargon.

Defining the Responses of Organisms. Since we have maintained
that the student of behavior begins by observing the actions or
responses of organisms, wt must next consider the problem of
how to specify or define a response. While difficulties may arise
at the conceptual level, in actual practice few problems accompany
the delineation of events that will be called responses. Responses
are whatever movements or classes of movements or actions we
decide to record or measure. A response is a member of a class or
set of objectively recordable events attributable to whatever organ-
ism we are observing. But if different observers are to agree that
a response has occurred (and such agreement is a necessary con-
dition for scientific analysis), they must all be governed by rules
of observation that specify precisely what activities are to be in-
cluded in a response class. For example, two observers will usually
agree that a rat has turned to the right at a designated point in
a maze if it turns sharply through an angle of 90°. But what if
the animal veers only 20° to the right of its original line of move-
ment? Is this a right-turning reaction? It is or is not a "right turn"
depending upon what limits have been placed upon the class of
movements to be designated right turns. The selection of particular
limits is a matter of choice at the outset, being governed by
hunches or prior knowledge. It is only when our response-defining
criteria have been shown to be unworkable or useless that they
must be revised. Thus one may decide to call a movement to
the right a right turn if it describes an angle of at least 80 but
not more than 100°. Once such class boundaries have been speci-
fied, any competent observer can decide whether a given turning
movement does or does not fall within the prescribed limits. If it
does, he makes a tally mark on his record sheet and the response
is said to have occurred.

From time to time various writers (Bakan, 1953; Nissen, 1954;
Hall and Lindzey, 1957) have suggested that information about
eliciting stimuli is essential to an adequate definition or description
of any response. This appears to be true only when the phrase
"adequate definition or description" is treated as though it were
synonymous with "adequate scientific explanation." Few will deny

4 , THE MOTIVATION OF BEHAVIOR

that every observable action of an organism is antedated by com-
plex chains of related events, some of which would unquestionably
qualify as "stimuli." Quite obviously, therefore, any complete
scientific explanation of a response would necessarily contain some
reference to correlated or eliciting stimuli.

But a scientific explanation of a response is one thing, and its
definition or description, for purposes of scientific observation, is
something else. The activities of organisms are physical events
occurring in the physical world, and agreement among competent
observers as to the occurrence or nonoccurrence of these actions
can be achieved through precisely formulated observational rules.
A response is adequately defined when the criteria for its observa-
tion have been specified in detail. For this purpose information
about the nature or presence of eliciting stimuli is completely
irrelevant. In making his raw tally marks, an investigator need
have no knowledge of the stimulus events that are correlated
with the behavior; it is required only that he know the rules of
observation and how to operate his measuring or recording equip-
ment. In fact, the behavior scientist can spend his entire life
investigating behavior and determining whether it is lawfully
related to other data without ever getting involved in the question
of whether he is indeed studying instances of what others might
call responses. Incidentally, the assertion that an action, a response,
or a reaction can be defined, specified, or described independently
of correlated stimuli does not imply that the behavior has no
stimulus correlates. It means simply that the behavior can be
identified and recorded even though its eliciting stimuli may never
be discovered.

Alternative Definitions of Response Classes. As we have seen,
the task of defining a response is not especially difficult provided
observational criteria can be specified with sufficient precision.
No two responses are ever exactly alike, however, and hence our
criteria of observation must encompass a class or group of re-
sponses. This necessity raises the further question of exactly where
the boundaries of the response class should be fixed. Prior to exten-
sive study we have no way of knowing whether the response
domain should be broad or narrow. Important laws might be
obtained with one criterion of class membership but not with

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 5

another. In the absence of information provided by experience,
the boundaries of response classes are determined by the best
available guesses as to which settings will prove most fruitful in
the long run. Psychologists' guesses in these matters have differed
widely, however, and considerable controversy has been the result.
Because of this difference of opinion it is necessary to examine
the problem in greater detail. This can best be done, perhaps,
by considering several of the alternative response-class definitions
that have been advocated by various investigators. Somewhat
similar discussions of response classes will be found elsewhere
(e.g., Davis, 1953; Logan, Olmsted, Rosner, Schwartz, and Stevens,
1955).

In the first place, some students of behavior are quite satis-
fied to work with a very limited class of reactions such as those
provided by a single muscle group or gland. Pavlov (1927), in
his extensive studies of salivary conditioning in dogs, used a
response-classification scheme of this kind. A dog was said to have
"behaved" or "responded" whenever a drop of saliva was seen to
fall from a tube attached to an externalized opening of its salivary
gland. Hilden (1937), in studying conditioned finger withdrawal
in human subjects, recorded electrical potentials originating in
the muscles controlling the finger. A "response" in this instance
was defined as an action-potential burst of a given magnitude.
Because of the method employed, a subject was said to be reacting
even when no overt movement of his finger could be detected.
If definite rules were prescribed for reading the graphic records
of electrical muscle activity, different observers would have little
difficulty in reaching an agreement as to whether a response had
or had not occurred. The psychologist who restricts the class of
recordable reactions to the activity of a single muscle group or
gland has adopted about as simple a classification scheme as he
can use and still maintain that he is studying "behavior." To
move farther back into the mechanisms underlying behavior, by
investigating, say, the neural events preceding the contraction of
a muscle, would take one into the traditional field of physiology.

A second, broader way of categorizing behavior would be to
limit one's observations to the overt movements of a single limb.
Thus one might tally the occurrence of a response, in an avoid-

6 THE MOTIVATION OF BEHAVIOR

ance-conditioning situation, if a dog were to lift its leg at least V4
inch from the floor during a certain interval following the presen-
tation of the conditioned stimulus. This way of categorizing
behavior differs from the first in that different muscle groups
might be involved on each occasion when the leg is lifted. But
by following the specified rules one can make repeated observa-
tions of the "same" behavior, where "sameness" is defined as
any upward movement of the foot of at least V4 inch. Such
repeated observations are, of course, essential in any scientific
investigation. The isolated occurrence of a single muscle twitch
or a single movement or a single act of aggression is of little
value to the student of behavior. Whether the criterion of move-
ment is set at V4 inch, 1 inch, or any other value is entirely up
to the psychologist who conducts the experiment.

A third way of classifying responses is to combine all activities
that have the same end effect upon the environment, regardless
of which specific limbs or muscles are involved. Thus, in studying
the behavior of children or adults, any action resulting in injury
to another person, regardless of how the injury was produced,
might be classified as an aggressive act. Or, in studying the behavior
of rats in a Skinner box, any depression of the bar might be treated
as a bar-pressing response irrespective of whether the bar is pressed
by the rat's foot, tail, head, or body. Different effectors and dif-
ferent kinds of neural integrations would be involved in each case,
but the reactions could all be treated as identical for the purpose
of counting responses. In every instance of this kind the, experi-
menter must specify the criteria to be met by the reaction if it is
to qualify as a recordable member of the response group under
study. If aggression is defined in terms of injury to others, then
the child who throws an object at his toy box, but misses and
hurts a second child, is exhibiting aggression. A bar-pressing
response of less than some predetermined amount may or may
not be included as a member of the class of bar-pressing acts. It
would not be included if it does not activate the food magazine
or the automatic electric counter that tallies the actions. Skinner
(1938) carries this procedure to its logical extreme. His experi-
mental situations are often structured so that it is impossible,
even should one so desire, to observe how the rat actually does

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 7

press the lever or to see whether partial responses are being made.
Skinner is satisfied to determine whether a response, defined and
identified in quite an arbitrary manner, varies lawfully with
changes in laboratory operations. If the response does not manifest
lawful variation, which might be the case if the force required to
depress the bar were markedly increased, then the response class
must be redefined and a new set of data collected.

Psychologists who observe in detail what an animal or child
does in a lever-pressing or problem-box situation may be impressed
by either the uniformity of behavior or by its variability. Guthrie
and Horton (1946), for example, in studying the behavior of
cats in a puzzle box, found their subjects' reactions to be highly
stereotyped. If the cat got out of the box by hitting the release
pole with its tail on one trial, it was likely to perform substantially
the same set of movements on the next trial. Others, such as
Tolman (1932), have been struck by the fact that, with regard
to the muscle groups involved, behavior is highly variable from
trial to trial. It is asserted, as a consequence, that the animal
is not learning to make responses mediated by activity of the
same efferent channels and the same muscular or glandular effec-
tors. Each of these points of view is, in essence, a statement of
faith as to how behavior should be classified if one is to attain to
the highest level of understanding, lawfulness, and predictive
efficiency. As will be noted shortly, those for whom the variability
of the underlying mechanisms looms large are impressed by the
constancy of the end result. Such writers often favor the view
that the behavior is "purposive." For them the animal seems to
know that regardless of which foot is used to press the lever the
food will be forthcoming.

A fourth way of classifying responses is to combine all actions
leading to the satisfaction of the same psychological or biological
need. Suppose, for example, that we have developed a method for
making reliable determinations of an individual's "need to be
aggressive." Conceivably, this need could be satisfied by physical
assault upon other persons or upon inanimate objects, by verbal
insults, by self-inflicted injury, and by a wide variety of other
actions, including imagined attacks upon others. If each of these
quite different types of behavior leads to the satisfaction of the

8 THE MOTIVATION OF BEHAVIOR

same need, they might all be included in a single response class.
Similarly, an -animal could satisfy its need for water by lapping
up any of several different kinds of fluid mixtures containing
water, by eating wet food, or by licking water from its paws.
Behavior of this sort is even more variable than that included
under our third method of classification, since it involves variations
in both effectors and goal objects. Because of this it is even more
likely to be called "purposive" by those who favor the use of such
a term.

As a fifth way of grouping actions into classes for purposes of
study, one might treat as class members all reactions involving
the same determinants or causes, regardless of the specific needs
satisfied, the environmental objects affected, or the limbs or muscle
groups involved. For instance, if one could decide unambiguously
that certain actions are controlled by the "will," and if other
observers could concur in these decisions, it might be useful to
allocate the responses to a "voluntary" category. Or if one could
ascertain which actions are governed by the organism's intentions,
these could be brought together into a class of purposive actions.
One might also be able to decide that certain activities are deter-
mined by personality traits such as anxiety, rigidity, or honesty
and thus be able to achieve useful groupings.

In concluding this discussion of criteria for classifying responses,
it should be noted that as we progress from the first through the
fifth method, it becomes increasingly difficult to get different
observers to agree that a given action has been correctly classified.
Actions can be defined independently of any potential or actual
stimulating conditions, simply by setting up criteria for counting
or measuring. But the criteria must be such that other impartial
observers can obtain similar results by applying the criteria under
essentially comparable conditions. Such application is not espe-
cially difficult with the first three methods, but it becomes a
relatively serious problem with the last two.

Incidentally, there is no reason why a single observer could not
apply more than one of these methods of response classification
to a single set of behavioral data. In making his observations, if
an experimenter has included a wide variety of actions and stimu-
lating conditions and if these have been carefully recorded, he

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 9

can, on later occasions, select several subsets of actions in accord-
ance with specific but different criteria. He is then free to utilize
whichever subset proves most useful or interesting.

Once criteria have been chosen for use in recording the occur-
rence of responses, the psychologist's next step is to search for
what may be loosely described as the "causes" of behavior. In
attempting to understand behavior it is not enough to record
behavior and nothing else. One could note, in the most minute
detail, the thousands of responses made by an organism through-
out its life and yet have little or no understanding of why the
behavior occurred. To understand behavior one must look for
environmental or organic conditions, for other responses of the
organism, or for any other events that can be related in some
meaningful way to the observed behavior. One must record events
or quantities other than the behavior under study: quantities such
as temperature, humidity, degree of illumination, sounds, physio-
logical states of the organism, number and kinds of previous
experiences, and, on many occasions, other reactions of the same
or different organisms. Because the task of defining and specifying
these additional events leads to some controversy among psychol-
ogists, we digress briefly to consider the problems involved in
defining the stimulus.

On the Definition of Stimuli. Some authors have maintained
that the stimuli for behavior can be defined only in terms of the
reactions they evoke. Because of this limitation they have argued
further that meaningful relations cannot be formulated between
responses and stimuli since the two variables of an empirical law
must be independent. From one point of view these are correct
assertions, but they must be carefully interpreted and extensively
qualified.

In the first place, by using the methods of the physical sciences,
we can specify or describe a multitude of physical phenomena in
an entirely objective manner. Moreover, this can be done quite
adequately without referring to the behavior of the particular
organisms we are investigating. We can measure the temperature
of the air that surrounds our subjects, the brightness of lights to
which they are told to respond, the intensity of sound produced
by a thunderclap, the number of grains of ragweed pollen per

10 THE MOTIVATION OF BEHAVIOR

cubic foot of air, the amplitude of radio waves, and so on. These
environmental events and many more are external to, and inde-
pendent of, the presence or behavior of the subjects of our psy-
chological experiments. For the moment let us describe these
phenomena by the term physical incidents.

We can also include in this category additional incidents at-
tributable to organisms other than our subjects. The patterns of
light and shadow provided by the movements of another organism,
the assorted sequences of sounds it produces, or its physical move-
ments, are phenomena that can be recorded and measured by
the same methods one uses for other physical incidents. These
incidents do originate in the behavior of an organism, but they
can be observed and measured independently of the particular
organisms that serve as our subjects. For some purposes it may
be desirable to distinguish this second group of incidents from
the first, but both kinds of physical occurrences can be measured
and defined by the methods of the natural scientist. Assertions
about the presence of any one of these incidents, about its dura-
tion in time, or about its intensity or extensity, depend solely upon
objective physical measurements; such assertions are not con-
tingent upon the presence of the organisms whose behavior is
under study, though they do depend upon the presence of the
observing and recording scientist.

In order to simplify this discussion we have ignored the possi-
bility that certain responses may have physical consequences that
can be correlated with subsequent responses. These response-gen-
erated physical events can, in principle, be measured in the same
ways that we measure events external to our subjects, but practical
techniques for doing so remain to be perfected.

A strong case can be made for the view, therefore, that the raw
data with which the psychologist works consist, in part, of (1)
a set of quantities obtained from physical measurements of the
actions of his subjects, and (2) a set of independent quantities
obtained by measuring or counting physical incidents originating
outside of his reacting subjects. The psychologist's goal is to under-
stand the first set of quantities, and to date the most fruitful
procedure has been to search for relations between elements of

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 1 1

the second set (the environment) and elements of the first (be-
havior). At a low level of complexity, the ability to state lawful
relations between the elements of these two sets of quantities
constitutes scientific explanation in psychology.

As the process of formulating and discovering dependable rela-
tions between these two sets of quantities progresses, the psychol-
ogist finds that some physical incidents are correlated with actions
of the observed organisms and some are not. For example, he
may note that a subject exhibits a vigorous startle reaction when-
ever a loud, sharp sound is presented. On the other hand, a rela-
tionship may never be found between intensity of cosmic radiation
reaching earth from outer space and any recordable bit of behavior.
Both the loud sound and the cosmic rays are physical incidents,
but, so far as we now know, only the sound is closely and con-
sistently related to behavior.

As a consequence of numerous observations, the psychologist
can point to a great many physical incidents that are related,
though in varying degrees, to behavior. Moreover, he frequently
knows how long the physical incidents must endure, how intense
and how close they must be to the organism, and what other
characteristics they must possess if they are to correlate with a
subject's behavior. On the basis of his investigations, therefore,
the psychologist can define a new subclass of physical incidents
which, though they differ radically from one another, have as a
common characteristic a known relatedness to actions. It is to
this subclass of physical incidents that the name stimuli is often
given. Thus, to describe a physical incident as a stimulus is simply
to assert that it is, as a consequence of empirical findings, known
to be related, or to have been related, to the behavior of some
organism in some way. The assertion that a stimulus must be
defined in terms of a response stems from the fact that responses
are involved in the process of attaching the name stimulus to a
given physical incident. But the inherent physical properties of
such a physical incident are not defined by appeal to responses,
but by reference to physical measurements. Whether a physical
event is to be called a stimulus thus depends on its being related
to responses; but what it is as a physical event does not. The term

12 THE MOTIVATION OF BEHAVIOR

stimulus is simply a convenient shorthand expression for the fact
of relatedness. It is a renaming that does little, in and of itself,
to advance our understanding of basic relations.

In connection with the general problem of deciding which
physical incidents might be chosen for inclusion in the special
class, stimuli, the following may be said: The responses used in
empirical demonstrations of relationships to physical incidents
may be of several different types. One may select any of the
response classes discussed above or may penetrate more deeply
into the physiological reactions of the organism. For example, one
may wish to assert that a response has been elicited by a physical
incident, which can then be called a stimulus if any afferent
activity whatever was noted when the incident occurred. If a faint
light leads to electrical activity in a subject's optic nerve, even
though no overt muscular movements occur, one may say the
light was a stimulus. Alternatively, one may choose to call the
light a stimulus only if its presentation results in the subject's
saying "I see something." In either event, reclassifying the light
as a stimulus is a consequence of observed concomitant action
exhibited by the subject. At present, there are apparently no
clear-cut reasons for choosing one kind of action over another.

Before turning to a further discussion of the task of the psy-
chologist, we should note that irrespective of which kind of activity
or response is used as the criterion for relabeling certain physical
incidents as stimuli, the relatedness itself is dependent upon a
great many variables whose presence must be considered. Thus a
light may qualify as a stimulus if it is brighter than a certain min-
imal value, if it is within a certain distance of the subject, if it is
located at a certain angle, if the subject is not too fatigued or too
bored, if response A is used rather than B, and so on. If one or
more of these conditions is not fulfilled, the light may not evoke
a measurable reaction, and if it does not, then at that moment it
does not meet the conditions for membership in the class, stimuli.
A thorough understanding of behavior requires that we know
precisely when and under what conditions a given physical incident
is indeed related to action. The task of discovering the nature of
these relations and the conditions under which they occur, con-

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 13

stitutes an extensive and important part of the psychologist's field
of endeavor.

Types of Functional Relations Studied by the Psychologist

As soon as the psychologist begins to observe both behavior
and the environmental circumstances within which it occurs, he
becomes involved in determining the empirical laws that hold
between and within the two sets of quantities. In the following
subsections we shall describe several kinds of these empirical laws.

Stimulus-Response Relations. Perhaps the simplest kinds of
empirical laws with which the psychologist is primarily concerned
are those in which the reactions of a subject (or subjects) to
systematically varied stimulus conditions are recorded. As an
illustration, suppose one were to measure the mean reaction time
of a group of subjects to each of several different intensities of a
visual signal provided by a light. If some regular change in reaction
time (R) were found to accompany the variations in light intensity
(S), a functional connection of the sort commonly called an S-R
law would have been obtained. Thus an S-R law is an empirical re-
lation between measured changes in some aspect of the physical
environment and measured variations in some property or charac-
teristic of responses, other factors being held constant.

An actual S-R relation obtained in a study by Hovland and
Riesen (1940) is shown in Fig. 1:1. As can be seen from this
figure, these investigators found that mean response amplitude
increased as the intensity of the tone was heightened. The response
was a decrease in skin resistance, and the stimulus was a 1,000-
cycle tone whose intensity above the subjects' thresholds was ex-
pressed in decibels.

In actual practice a function of this kind could be obtained
either from repeated tests of the same subjects, using the five
different stimulus values, or from tests of a different group of
subjects with each value. In either case the resulting S-R function
constitutes a summary of the results of all the tests. The S-R laws
obtained from individual subjects would probably differ consider-
ably from an averaged function like that of Fig. 1:1.

THE MOTIVATION OF BEHAVIOR
40

30 60 90 120

Sound intensity in decibels above threshold

Fig. 1:1. A simple S-R relation, in which the amplitude of a galvanic skin
response (GSR) becomes progressively greater as the intensity of a sound
stimulus is increased. (Adapted from Hovland and Riesen, 1940.)

R-R Relations. A second kind of empirical law frequently ob-
tained by psychologists may be described, following Spence (1944),
as an R-R relation. Such a law is obtained by plotting one set of
response measures against another. Suppose we administer an
arithmetic aptitude test to three college students, who are required
to solve as many addition problems as possible in five minutes.
Suppose further that student A solves 20 problems, student B
solves 15, and student C solves 10, These scores constitute one
measure of the responses of each student to the stimuli provided
by the test situation. Now imagine that the same test is given
to the same three students a day later, and the scores of 23, 18,
and 13 are obtained by A, B, and C, respectively. The results of
these two administrations of the same test, if plotted graphically,
would look like Fig, 1:2. This constitutes an R-R law of an ex-
tremely simple and elementary sort.

Nevertheless, this law contains the essential ingredients for
making predictions of a relatively primitive variety. From a knowl-
edge of how a student performs on the first day's test we can
predict, with a certain margin of error, how well he will do on

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION

22 -

-S20 -

18 -

"Si 16 -

•2 14-

Student A

Student B

Student C

15
Addition test score — day one

Fig. 1:2. Hypothetical R-R relation obtained by administering the same
addition test to the same subjects on each of two successive days.

the second day. This assumes, of course, that the relationship
between the performances on days one and two is essentially
stable. To the student acquainted with elementary test-construc-
tion methods it will be apparent that this simple R-R law is iden-
tical in form to the relation obtained when one determines the
test-retest reliability of a test. If an individual gets approximately
the same absolute or relative score upon repeated administrations
of the same test, or comparable forms thereof, the test is said to be
reliable. It measures the same capacities or abilities each time
it is administered.

Different K-K laws will of course be obtained if the two re-
sponses are measured under different environmental conditions
and by different tests. To illustrate, the three students of the
preceding example might all have taken the same elementary
mathematics course, and student A might have received a grade
of 90, student B a grade of 85, and C a grade of 70. When these
course grades are plotted against the first day's addition-test scores
for the same students, the resulting R-R law might be of the form
shown in Fig. 1:3.

An R-R law such as this, assuming again that it proves to be
stable, has considerable utility. From a knowledge of a student's
score on the addition test one can predict his course grade, and

THE MOTIVATION OF BEHAVIOR

Student A

^^-"''''^

•o

^^^^

^ 85

•'''^

/Student B

w 80

£ 75

- *^tudent C

1 1

15
Addition test score

Fig. 1:3. Hypothetical R-R relation between the scores obtained by three
students on a simple addition test and the grades they received in an ele-
mentary mathematics course.

vice versa. Readers familiar with elementary statistics will imme-
diately recognize this relation as a very simple scattergram like
those used in connection with the computation of coefficients of
correlation. B<-B^ relationships like this are extensively employed
by psychologists who work in the areas of vocational guidance,
personnel selection, and industrial psychology. If one can success-
fully predict, from the results of a short, written test, how well
an employee will do in a certain job or how well a student will
perform in college, substantial savings in time and money can be
effected.

These two examples of K-K laws may be thought of as lying
at the extremes of a continuum. At one end the same test items
are presented on two separate occasions in the same environment,
and nearly identical responses are recorded. At the other extreme
neither the responses being recorded nor the two test situations are
markedly similar. By making the test conditions and/or the
responses the same or different we can achieve a wide variety of
K-K laws that may be placed at appropriate positions on this
continuum. The same kinds of responses might be recorded on
the two occasions, but the testing situations might be quite differ-
ent. Or the situations might be identical, but different reactions

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION

might be recorded. When both the reactions and the testing
situations differ, as in our second example, the resultant R-R
function is like those obtained in the process of validating a test.
O-R Relationships. A third type of function that psychologists
seek to establish may be described as an OR law (Spence, 1944).
This designation refers to a relation in which the independent
variable is some measurable organic characteristic, property, or
state of an organism {or of different groups of organisms), and
the dependent variable is some reaction or response. An OR law
differs from an S-R law in that the independent variable in the S-R
relationship is usually a physical event in the external environment,
whereas in the OR law it is some bodily state of an organism. An
example of an OR law is given in Fig. 1 :4 where the R variable
is the number of times female rats crossed an electrified grid to
reach a male and the O variable is the stage of the estral cycle
as defined by histological examinations of the vaginal mucosa.
In this experiment (Warner, 1927) all of the subjects were first
tested in the obstruction box and were then classified into separate
groups according to the results of the histological tests. In a sense,

ra ^ oi2

Stages of estral cycle

Fig. 1:4. An example of an OR law. The organic variable (O) plotted on
the base line is the stage of the female estral cycle, and the response variable
(R) on the ordinate is the number of times the female rats crossed an elec-
trified grid to reach a male. (Adapted from Warner, 1927.)

18 THE MOTIVATION OF BEHAVIOR

estrous level was thus varied by choosing subgroups of subjects
who fell at each stage.

Specific values of an O variable are usually determined by means
of standardized physical methods of measurement. For instance,
the basic method used to compute percentage of alcohol in the
blood of automobile drivers charged with intoxication probably
involves chemical analyses of blood samples. Under certain condi-
tions, however, comparable results could be obtained from be-
havioral data. Thus if a reliable functional connection (empirical
law) has been established between alcohol percentage and scores
on a line-walking test, then O values can be estimated from scores
on the behavioral test.

Other Varieties of Functional Relations. The S-R, R-R, and OR
relations discussed above are among the most common empirical
functions obtained by the research psychologist. There are several
others, however, which, because of the ways we have chosen to
apply our labels, do not qualify precisely as S-R, R-R, or OR func-
tions. For instance, the familiar learning curve relating some
measure of performance to number of trials has not been cited
here as an example of an S-R law, though it is often so regarded,
since the physically defined properties of the stimuli in the learn-
ing situation are not systematically varied over trials but are held
as constant as possible. Likewise, relations in which behavioral
changes are plotted against the passage of time, e.g., speed of
movement as a function of chronological age and sensitivity to
light as a function of time in darkness, cannot be subsumed easily
under any of the types of laws already considered. One could
readily invent new alphabetical abbreviations for laws such as
these, but the advantages to be gained therefrom appear to be
slight.

Although in each of the functions described above a response
measure was plotted on the ordinate of the graph, this is not true
of all psychological laws. For example, the psychologist might be
interested in determining the visual thresholds of his subjects
under conditions where both the duration and the brightness of
a test light are systematically varied. The empirical function
resulting from such an experiment might look like the graph
in Fig. 1:5. Here the intensity of the test light is plotted along

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION
0.03
0.02

S 0.01 h

0.001

2 4

Stimulus intensity

Fig. 1.5. An empirical relation in which physically measured values of the
stimulus are plotted on both dimensions. Each point on the curve represents
a visual stimulus whose duration and intensity are such that it is seen just
50 per cent of the time. {Adapted from Braunstein, 1923.)

the abscissa and its duration on the ordinate. The empirical
points tell us which combinations of duration and intensity result
in the subject's seeing the test light 50 per cent of the times on
which it is presented. Thus the light will be seen for half the time
if it is brief and bright, or for half the time, even when it is dim,
if it is presented for a longer time. This type of relation differs
from the S-R variety in that physically measured quantities occupy
both the ordinate and the abscissa. Moreover, the plotted points
do not represent different magnitudes or frequencies of a response.
Because of these differences, this particular sort of function might
best be described as a modified S-R relation.

Finally, it should be noted that functions more complicated
than either the simple S-R or OR laws can be obtained by simul-
taneously varying both the condition of the organism and some
characteristic of the stimulus situation. Relations obtained under
such conditions might be termed SO-R laws, since the dependent
variable (the response) is a joint function of both an organic
variable and an environmental one. A hypothetical example of
this kind of relation is presented in Fig. 1:6. In preparing this
graph it has been assumed that reaction time decreases as a

THE MOTIVATION OF BEHAVIOR

Hypothetical

levels of

ascending

reticular

system

activity

Stimulus intensity

Fig. 1:6. Fictitious data depicting SO-R relations. The dependent variable
(reaction time) is shown as varying both with stimulus intensity and amount
of general neural activity in the ascending reticular activating system.

function of increasing stimulus intensity but that the function
relating the two variables takes a different form depending upon
the average level of activity of the brain-stem reticular formation.
In this case stimulus intensity is the principal independent variable,
and reticular-system activity, though also classifiable as an in-
dependent variable, would usually be described as a parameter.
This combination-type law exemplifies a phenomenon which
statisticians describe as an interaction. By this they mean that the
effect of stimulus intensity upon reaction time depends upon the
particular level of the third factor, reticular-system discharges. Al-
though the curves of Fig. 1:6 have been called an SO-R relation,
three simple and distinct laws are actually plotted on one graph.
Each law, considered alone, is an S-R law specific to the magnitude
of neural excitation present at the time of its determination. Level
of reticular outflow could, with equal justification, have been
plotted on the base line as the major independent variable, with
different values of stimulus intensity constituting the parameter.
To maintain a consistent terminology, compound laws of this
latter type should probably be termed OS-R relations.

The reader should note that if the three values plotted vertically
at any point on the abscissa of Fig. 1 :6 were to be replotted as a

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 21

separate graph, the result would be an OR law relating reaction
time to three values of the independent O variable (reticular
activity), with stimulus intensity held constant. Similarly, a re-
plotting of the values at any abscissa point on an OS-R relationship
would yield an S-R law, the value of O being held constant. The
complexity of any of these relationships could be increased by
introducing other factors as additional parameters. The two-
dimensional curves of Fig. 1 :6 could be expanded into three-dimen-
sional surfaces by simultaneously varying some other factor in
addition to stimulus intensity and reticular outflow. The number
of variables that might be included is limited only by the experi-
menter's ability to control and measure them and to analyze and
comprehend the results.

General Approaches to the Study of Behavior

As noted in our introductory paragraphs, the psychologist's fun-
damental task is to study and to attempt to understand the be-
havior of organisms. Although some progress toward this goal
might be attained by a careful study of historical and literary
writings, most students of behavior believe that primary reliance
should be placed upon objective observations of the activities of
organisms and of environmental or other factors that are corre-
lated with such actions. When the data obtained from such
observations are collated and organized, the results are empirical
relations of the several varieties we have just described.

But no matter how many functional relations are established
by the psychologist, he is seldom satisfied to cease his inquiries at
that point. He may be pleased to learn that reaction time decreases
as a function of increasing stimulus intensity, but he is even more
pleased if he can gain additional insights into the factors that
determine this function. Hence he strives to learn precisely why
reaction time declines as stimulus intensity increases. In his efforts
to understand behavior, therefore, he often finds it rewarding to
go beyond his initial empirical data, to speculate about relations
not yet discovered, and to search for new facts and new contin-
gencies.

Given a well-established empirical law, the psychologist may

22 THE MOTIVATION OF BEHAVIOR

follow two principal paths toward the enrichment of his under-
standing of the law. One of these, which involves detailed analyses
and descriptions of the bodily mechanisms of behaving organisms,
is usually described as the "physiological approach." The second,
often characterized as the "behavioral approach," leads to further
study of the range of conditions under which the law holds and
to the search for "explanatory" laws in which only behavioral
and experimental variables are contained. These two approaches
are not mutually exclusive, and many investigators follow both,
but they are suflBciently different to warrant separate discussion.

The Physiological Approach. The student of behavior who
adopts this general plan of action tries, where possible, to interpret
his empirical laws by appealing to known facts concerning an
organism's physical structure and its functions. He endeavors to
explain his findings in terms of knowledge and concepts derived
from the work of the physiologist, the physiological psychologist,
the anatomist, the biochemist, and the biophysicist. Given an
empirical relation between a response measure and a stimulus
variable, he may inquire into the details of the receptor mechan-
isms by which the organism senses the stimulus, or he may try to
trace the neural connections intervening between the receptor
and the final response. This approach is primarily favored by
those who concentrate on the study of the sensory processes of
vision, audition, olfaction, and gustation. Its value is indicated by
the tremendous strides toward the understanding of such processes
that have been made during the past few decades. And the desir-
ability of continued work along these lines is strongly supported
by the widespread conviction among behavior scientists that living
organisms, though descriptively unique and astonishingly complex,
are nevertheless physical systems or assemblies of such systems.
There are many reasons for believing, therefore, that in the future
our explanations of complex human and animal behavior will be
considerably broadened by the identification of the neurophys-
iological and physicochemical systems whose functions make that
behavior possible.

The Behavioral Approach. Investigators whose activities exem-
plify the behavioral approach typically do not concern themselves
with the question of why an empirical law is what it is in the

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 23

light of the structural characteristics of the behaving organism.
They do not, as a rule, apply the terms or concepts of the physiol-
ogist or anatomist to the explication of empirical relations between
behavior and other variables and seldom refer to such specific
bodily entities as muscles, tendons, glands, and nerve fibers.
Instead, their attention is concentrated upon what they may
describe as molar behavior, meaning relatively gross movements
or goal-oriented actions of the entire organism. For the psychol-
ogist who favors this approach reasonably satisfactory explanations
of behavior can be wrung from a knowledge of the wide assort-
ment of experimental variables that affect behavioral laws, from
information as to the kinds and magnitudes of these effects, and
from supraordinate laws that tie simpler laws together in a mean-
ingful fashion. Gaps in this network of explanatory laws are filled,
in part, by "guessed-at laws" (Spence, 1948) and by the introduc-
tion of "explanatory" concepts such as intelligence, personality
traits, ego-involvement, associative connections, inherited reactive
tendencies, unconscious desires, cognitions, motives, and drives.
These terms refer, though often in a relatively vague fashion, to
properties, states, predispositions, or characteristics of organisms
that function as determinants of behavior. Their basic meanings,
for purposes of communication within the context of the scientific
vocabulary, can be given by explicit definitions, but failure to
provide such definitions is commonplace. The broader meanings
of these terms, in their roles as integrative or explanatory elements
of interpretive networks or theories, derive from the varieties and
kinds of empirical and conceptual relations into which they enter
as constituents. In following the behavioral approach, therefore,
one deals primarily with molar rather than molecular behavior;
one investigates the ways in which an assortment of variables
affects behavior and modifies the laws relating it to other variables;
and in attempting to explain behavior, one tends to appeal to pos-
tulated processes or intervening variables whose principal function
is that of summarization and integration. The concepts of drive
and motivation, to which we now turn, are often encountered in
the writings of those who subscribe to the behavioral approach,
and generally speaking, occupy a position of considerable im-
portance therein.

24 THE MOTIVATION OF BEHAVIOR

The Problem of Motivation

There is no question but that the idea of motivation or some
similar notion appears in almost every systematic account of
behavior. Contemporary psychological theorists as well as their
more philosophically oriented predecessors have frequently relied
upon some kind of moving, pushing, driving, or energizing force
or agency. The ubiquity of the concept of motivation, in one
guise or another, is nevertheless surprising when we consider that
its meaning is often scandalously vague. It is not our intention to
tabulate all of the many motivationlike terms that have been used
or to summarize the history and development of the concept. Such
summaries may be found in the writings of Troland (1928),
Young (1936), Gardiner, Metcalf, and Beebe-Center (1937),
Lindzey (1959), and Madsen (1959), as well as in widely scattered
briefer discussions. It will be sufficient to note that, depending
upon the particular writer consulted, motivation can be conscious
or unconscious; it can be the same as, or different from, drive;
it may or may not guide behavior; and all motives can be either
learned or instinctive. Moreover, arguments can be found to sup-
port the view that motivation is both crucial to behavior and a
useless concept, that it is simply the energy that moves the body,
or that it is identical with the neural discharges of specific central
nervous-system structures. We thus find ourselves in the position
of trying to deal with an allegedly vital factor in the face of violent
disagreements as to its origins, its essential nature, and its particu-
lar roles as a behavior determinant. The evaluation and reconcilia-
tion, where possible, of these divergent opinions and contradictory
views is of central concern to the student of motivation and is
a significant facet of the problem of motivation.

Within the field of psychology, broadly defined, factors or vari-
ables known to affect behavior in one way or another are grouped
into a number of loosely defined classes, to each of which, by
more or less common agreement, a distinguishing name is assigned.
These names frequently coincide with the traditional chapter
headings of our elementary texts. For instance, when behavior is
found to vary with changes in sensory stimulation, the psychologist
says he is concerned with the problems of sensation or perception.

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 25

If frequency and diversity of previous experiences prove to be
important variables, the area of study may be labeled learning or
fatigue or adaptation, depending on other information. If verbally
administered instructions influence behavior, the research is con-
cerned with set or expectancy. In like manner, other areas of
investigation are given names such as conflict, emotion, transfer
of training, remembering, and of course, motivation or drive.

Although significant and relevant variables have been success-
fully isolated and identified in some of these areas, this is less true
for motivation than one might wish. Many of the fundamental
problems in this field arise when serious inquiries are launched
into the nature of motivational variables. To bring order into
our thinking we need criteria for deciding that a given variable
is indeed affecting behavior "motivationally"; we need to know
whether motivational variables can be identified in terms of in-
trinsic properties as well as by means of their effects on behavior;
and we need to identify variables that may function both motiva-
tionally and nonmotivationally. The task of obtaining such knowl-
edge is clearly relevant to the general motivational problem, and
much of the remainder of this book is concerned with this matter.

It is also necessary to note that in recent years more and more
investigators have raised the serious question of whether a concept
of motivation is really required by a comprehensive theory of
behavior. The student of motivation must also devote a portion
of his time, therefore, to the analysis and evaluation of arguments
and evidence bearing on this issue. This means that he must con-
cern himself not only with the problem of motivation but also
with more generally comprehensive behavior theories. The con-
struction and evaluation of theories in which motivation plays
an important role is as much a part of the general problem of
motivation as the gathering of empirical data and the identification
of significant variables.

Summary

In this chapter we have tried to set the stage for our subsequent
discussions of motivation by describing in general terms some of
the activities, procedures, and goals of the student of behavior.

26 THE MOTIVATION OF BEHAVIOR

The attainment of satisfactory explanations of behavior is described
as the psychologist's principal aim and the observing and recording
of the activities of organisms as his initial task. The raw data with
which he works consist of tabulations of the frequency of occur-
rence of certain actions, their magnitude, their rate, or their dura-
tion. These data are unique in that they are contingent upon the
presence of living organisms, but the methods used in collecting
them are essentially identical with those of the natural sciences.

Since no two responses are ever exactly alike, the investigator,
in gathering his basic data, must deal with classes or groups of
responses. The boundaries of the class may be quite restricted, as
in the case where the activity of only a single muscle group or
gland is recorded, or wide enough to encompass complex purposive
and goal-directed acts. The setting of class limits is dictated solely
by individual preference and by the scientific and/or practical
utility of the research findings. Irrespective of where the boundaries
are set, however, lawful regularities in behavior cannot be expected
unless the conditions of observation have been specified precisely
in advance.

Collections of raw facts, valuable as they are, do not contain all
of the ingredients required for adequate scientific explanations.
Unelaborated descriptions of a subject's behavior must be sup-
plemented by accounts of environmental events, by records of
previous experiences and reactions, by observations of the behavior
of other organisms, and by measurements of physiologically de-
fined states or characteristics of the subject. The discovery of
dependable empirical laws relating behavior to other variables
such as these is a necessary step toward explanation and under-
standing. Identifying abbreviations such as S-R, R-R, and O-R are
often applied to special groups of these laws.

Because the component variables of any empirical relation must
be independent of one another, it is sometimes held that one
cannot legitimately speak of stimulus-response laws, since it is
also alleged that responses and stimuli can be defined only in
terms of one another. An analysis of this matter leads to the
conclusion that, given precise criteria of observation, responses
can be identified and recorded reliably even when observers know
nothing of the stimuli that elicit the responses. Similarly, the

PSYCHOLOGIST'S TASK AND PROBLEM OF MOTIVATION 27-

conclusion is reached that environmental events such as hghts,
sounds, and odors can be identified and measured by physical
procedures, irrespective of whether these events are correlated with
responses. To apply the term stimuli to events having known
response contingencies is simply to reassert the facts of dependency.

The discovery of empirical relations represents an important
step toward scientific explanation, but the questions of why a
given law holds or why it takes a particular form also demand
answers. In seeking solutions some investigators find it profitable
to study the physiological and neurological bases of behavior and
may be said to follow the physiological approach. Others, in adopt-
ing a behavioral approach, confine their search for answers to data
at the behavioral level of description. Their techniques involve
the discovery and evaluation of further related laws, the manipula-
tion of additional variables to determine the range of factors
crucial to the stability of laws, and the introduction of summariz-
ing or explanatory concepts such as habit strength, personality
traits, intelligence, inhibition, and motivation. The meanings of
these terms are given both by explicit definitions within the scien-
tific language and by the variety of useful relations into which
they enter as constituents.

Finally, although a concept of motivation or some similar notion
is to be found in nearly ever}' theoretical account of behavior, an
amazing divergence of opinion exists as to the nature and function
of motivation. A significant portion of the general problem of
motivation arises, therefore, from the need to clarify, evaluate,
and, wherever possible, to reconcile these disparate conceptions.
Other critical aspects of the problem are the identification, selec-
tion, and manipulation of motivational variables, the critical ap-
praisal of the utility of the concept, and the formulation of
adequate theories.

CHAPTER

Intervening Variables and
the Definition and
Measurement of Drive

As WE HAVE SEEN in the previous chapter, investigators who at-
tempt to formulate systematic accounts of behavior often appeal
to intermediary explanatory factors such as intelligence, personality,
field forces, associative strength, libido, instinct, frustration, and
drive. Various general names have been suggested for these inter-
mediaries, e.g., symbolic constructs, explanatory concepts, and
hypothetical constructs, but the term intervening variables (Tol-
man, 1932) is perhaps the most common. The qualifying adjective
"intervening" is used to convey the notion that the postulated
states, conditions, or processes intervene between behavior and its
observable correlates or antecedents. Since these variables cannot
be observed directly, their meanings are provided by explicit defini-
tions and by their functional relations within the context of
general theories of behavior. Inasmuch as the introduction of such
conceptions is widespread, and since motivation is a prominent
example, it is desirable at this point to consider the question of
why students of behavior have felt impelled to make venturesome

INTERVENING VARIABLES AND DEFINITION OF DRIVE 29

and sometimes unbridled speculations about such unseen entities.
One rather obvious reason for the use of intervening variables,
even when ill defined and loosely connected to a theory, is that
the users regard them as having real value as summarizing or
interpretive concepts. The psychologist feels that he has explained
behavior in a way that would otherwise have been impossible.
Alternatively, some explanatory concepts ( and certainly motivation
is one of these) are elements of the nonscientific language of the
layman and may have been passed uncritically from one generation
of psychologists to another without regard for their scientific
worth. One cannot, unfortunately, ask the psychologists of the
last fifty years why they have felt the need to incorporate inter-
vening variables into their theories. We can, however, examine
the kinds of data with which they have worked to determine
whether those data or their interrelations have special character-
istics requiring the postulation of intermediary concepts. It is
difficult to believe that the universality of such notions as asso-
ciative strength, drive, and the like, can be due merely to chance,
and it is possible, therefore, that intervening variables have, in a
sense, been thrust upon psychologists by the nature of the em-
pirical facts with which they have dealt.

Empirical Relations Tending to Evoke Motivational
or Other Intermediary Concepts

The Evocation of Vigorous Responses by Weak Stimuli. Con-
cepts such as drive, set, or expectancy are especially likely to
appeal to psychologists when subjects exhibit violent reactions to
weak stimuli. The creaking of a wind-blown shutter may evoke
intense reactions of fear or escape from the timid explorer of an
abandoned house; the faint nighttime stirrings of a sick child may
galvanize its mother into action; and the smell of a female dog
in heat may rouse the male to excited and extensive exploratory
action.

Disproportionalities of this sort are common among S-R laws,
and though they may assume a variety of forms, a single example
will suffice. In this instance, the relation is such that increases in
the intensity of the stimulus produce no response whatever until

THE MOTIVATION OF BEHAVIOR

100

o) 50 -

12 3 4 5 6 7 8

Stimulus intensity

Fig. 2:1. Diagrammatic representation of a stepwise disproportion between
the intensity of a stimulus and the strength of the response it evokes. The
units on both axes are fictitious.

a certain level is reached, whereupon the reaction appears with
maximum strength. Further increases in the strength of the stimu-
lus have no effect upon the response. A stepwise function of this
type is shown in Fig. 2:1, where the intensity of the stimulus has
been plotted on the horizontal axis (abscissa) and the vigor of
reaction on the vertical axis (ordinate). The units of both scales
are fictitious. A function of this type might be obtained in the
case of verbally presented instructions or commands. If the instruc-
tions are too faint to be heard, the receiving organism does noth-
ing; but when they become loud enough to be clearly understood,
the hearer reacts appropriately. Beyond the point of intelligibility,
further increases in the intensity of the commands may have little
or no effect on the behavior they elicit.

In dealing with a relation of this sort, the psychologist may
entertain the view that the stimulus controls a separate source
of energy which is released in the response, and hence may find
it desirable to introduce a motivationlike concept. It must be
emphasized, however, that disproportionality, as such, can also be
explained, perhaps more adequately, by nonmotivational concepts
such as set, expectancy, or associative strength. Motivation may be
suggested by data such as these, but it is not required.

Variability of Response in the Presence of Constant Stimulating
Conditions. The observation that an individual's behavior varies

INTERVENING VARIABLES AND DEFINITION OF DRIVE

from time to time in an environment whose physical characteristics
have not apparently changed is commonplace and provides strong
inducement for the use of one or another intervening concept.
Suppose, for example, that the running speed of a rat is measured
in the same straight runway on three different occasions and that
at each successive trial his speed increases. The elements of this
situation are represented in Fig. 2:2, with the large S at the left
of the diagram denoting the constant environment provided by
the alley and the three Rs at the right, with their respective sub-
scripts, indicating the running speeds as measured on the three
trials.

Presented with data of this kind, the investigator may wonder
why the rat did not run at exactly the same speed on the three
occasions. Since the physical properties of the runway were iden-
tical throughout the experiment and the rat was presumably
handled in the same way on each occasion, performance inequali-
ties cannot be explained by appealing to variations in the stimulus
situation. At this point the psychologist might assume that the
rat differed from trial to trial with respect to some unknown
state or characteristic, and that this characteristic was responsible
for the recorded changes in running speed. As a first step toward
accounting for his observations, therefore, he might postulate the
existence of a behavior-determining fac-
tor which he simply calls factor X. This
leads him to modify the relations in Fig,
2:2 to include three values of X. This
alteration is shown in Fig. 2:3, where
the greatest amount or degree of X, i.e,,
Xhigh, is associated with the highest
speed (Rfast), the intermediate level of

X with Rmoderate, and loW X with Rsiow

From this, the scientist may wish to as-
sert that speed of running depends upon
level or amount of factor X,

Not much has been accomplished, of
course, if the interpretive process stops
at this point. A conception as bare as
this must be strengthened and enriched

Fig. 2:2. Schematic illus-
tration of intraindividual
variation in response de-
spite a carefully maintained
constancy of the external
environment (S). The sub-
scripts to the three capital
Rs indicate the running
speeds of a single rat when
tested on three occasions
in the same straight-alley
maze.

THE MOTIVATION OF BEHAVIOR

^R.

Fig. 2:3. Schematic diagram show-
ing the postulated introduction of
three levels of a hypothetical be-
havior determinant (X) designed
to "explain" differences in the re-
sponses of a single organism that
has been tested under identical
stimulating conditions (S) on three
occasions.

by additional assumptions as to the properties of factor X, the pre-
cise ways in which it functions to determine speed of running, and
its relations to experimental variables and other intervening con-
cepts. These are matters of substantial importance to the behavior
scientist, but their consideration must be postponed for the pres-
ent.

Enlarging upon this example of response variation in the pres-
ence of constant stimulus conditions, we note that the relation,
in its basic form, applies to differences between the performances
of different individuals as well as to intraindividual variability.
Thus if one rat runs faster than a second, and the second faster
than a third, when all are tested in the same situation, an inter-
vening, performance-determining variable can perhaps be profitably
invoked.

An instance of what seems to qualify as a special case of response
variability in the presence of constant stimulation is found in
the observation that, when food is presented to an animal, it some-
times eats and sometimes does not; or, of two supposedly identical
animals, one may eat and the other not. Skinner (1938), in his
treatment of motivation, states that this is the basic phenomenon
giving rise to the concept of drive. This particular behavior could
be represented by separate diagrams like those of Figs. 2:2 and 2:3,
but this is not necessary. Suppose, for example, that the rat whose
performance is shown in Fig. 2:2 instead of running slowly, simply
does not run at all. His failure to run might then be attributed to
a complete lack of, or negligible strength of, factor X. If the
behavior is exhibited, X is postulated to be present; if no behavior
occurs, X is assumed to be effectively absent.

Equality or Constancy of Behavior in the Presence of Normally
Effective Changes in the External Stimulus Situation. Under some

INTERVENING VARIABLES AND DEFINITION OF DRIVE 33

Fig. 2:4. Three different stimulus
situations, i.e., the brightnesses of
an alley on three different occasions,
are represented at the left by the
three Ss. It has been assumed that o . — >.p ^ .

"^bright ^•"•moderate

these unequal brightnesses all lead
to equal (and moderate) rates of

running in three individual rats, Smoderate ^-^ moderate

even though rats usually run faster
(also assumed) under bright than
under dim illumination. Sji,^ ^-Rmoderate

conditions subjects may behave alike even though they are ex-
pected to react quite differently, and an individual, in situations
that typically lead to quite different responses, may show behav-
ioral constancy. As in our preceding examples, empirical data of
this sort tend to lead to the introduction of intervening variables.
To give a specific illustration, imagine that, normally, the more
brightly an alley maze is lighted the faster a rat will run. Suppose
further that an available alley can be illuminated with three de-
grees of brightness and that three rats from the same litter are
tested therein, one under each level of illumination. If all three
animals run at the same speed, we would have an instance of
response equality in the presence of normally effective changes in
the external situation. This is represented schematically in Fig. 2:4.

Were such results obtained, the psychologist might be tempted
to explain them by assuming that the three rats differed with
respect to a factor X. He might postulate that the rat tested under
bright illumination ran more slowly than animals typically do
because its level of X was low. Similarly, the rat tested under
dim illumination might have run faster than expected because of
an excessive amount of X. By assuming that differences in X have
counteracted the effects of illumination in this way one might
explain the failure to obtain the expected differences in response.
The elements of this interpretation are represented diagrammat-
ically in Fig. 2:5.

In concluding this section, we must note that many additional
examples might be cited of representative empirical relations that
seem to call for the introduction of intermediary constructs. For
example, as Skinner (1938), Hull (1943), and Miller (1959),
among others, have observed, when several different experimental

34 THE MOTIVATION OF BEHAVIOR

^bright -^low ^-"n

-^Rr

■^dim high

X..„. ^Rr

Fig. 2:5. Showing how equahty of response in spite of normally effective
variations in the brightness of a stimulus might be explained by postulating
differing levels of an intervening variable, factor X. Variations in brightness
and variations in factor X are inversely related when evaluated with respect
to their effects on speed of running.

variables are found to affect behavior in the same way, the postu-
lation of an intervening process common to all of them may prove
desirable. Furthermore, intervening constructs should not be re-
garded as unique to S-R relations. Equally imperative reasons for
the use of such concepts may arise from the study of OR, R-R,
and other relations.

Circular Reasoning in the Use of intermediary Concepts

When properly introduced into a psychological theory, interven-
ing variables often serve a useful function as conceptual devices.
When carelessly defined and irresponsibly used, however, they not
only contribute nothing to our understanding but may even be
gravely misleading. In the preceding section, several examples
were given of the kinds of empirical relations that seem to give
rise to intermediary concepts, and it was observed that in attempt-
ing to explain such relations one might begin by attributing the
behavior to variations in an unidentified factor X. One must go
well beyond this point, however, to achieve satisfactory explana-
tions. It is especially unfruitful to introduce different degrees of
factor X on the ground that different responses are observed, and
then to turn about and appeal to the inferred values of X to "ex-
plain" the variations in behavior. One may assume that some ani-
mals run faster than others because they have more of factor X,
but this is trivially circular if the only basis one has for asserting
they have more X is the fact that they run more rapidly. From
this it becomes clear that the introduction of factor X in the man-

INTERVENING VARIABLES AND DEFINITION OF DRIVE 35

ner diagramed in Figs. 2:3 and 2:5 has in only the most superficial
sense explained the observed behavior.

There are many instances within psychology of circular reason-
ing in the use of intermediary constructs. For instance, one fre-
quently encounters the statement that individuals behave in differ-
ent ways in the same environment because their perceptions are
not the same. This reasoning is typically circular since the percep-
tions are defined by the same behavior that is allegedly determined
by the perceptions. Other terms such as ability, constitutional dif-
ferences, personality traits, and motivation are likewise often in-
troduced and used in this uncritical way. Psychology is not alone,
however, in its occasional use of circular arguments. Social scien-
tists, for example, have tried to explain the gregarious behavior of
sheep by postulating an instinct to be gregarious (the herding
instinct), but the existence of the instinct has usually been based
entirely upon the observed gregarious behavior.

Defining Intervening Variables so as
to Avoid Circular Explanations

As we have seen, the introduction of intermediary concepts into
one's theory must be done in such a fashion as to avoid circular
interpretations. This can be accomplished, in principle, by making
certain that one's definition of an intervening variable is com-
pletely independent of the specific responses that are assumed to
be determined by that variable. This requirement can be met by
basing the definition upon ( 1 ) the subject's previous experiences,
(2) the responses he makes in other test situations, (3) one or
another of his organic states, or (4) the stimuli impinging upon
him. Each of these ways of achieving independence in definition
is examined in more detail in the paragraphs that follow, and in
every case the intervening variable is simply called factor X. The
problem of whether factor X should be described as drive, habit
strength, cognition, set, or whatever, will be discussed at a later
point in this chapter.

1. An Intervening Variable May Be Defined in Terms of Dif-
ferences in Experience Prior to the Test Situation. In essence, this
method of defining an intervening concept requires that we have

36 THE MOTIVATION OF BEHAVIOR

some knowledge concerning an organism's life history. For in-
stance, if subjects who respond differently when tested in the same
environmental situation can be shown to have had different ex-
periences or treatments prior to the test, then those experiences
can be used to define a factor X.

To illustrate, suppose we have carried out an experiment like
that depicted in Fig. 2:2, using three rats rather than one, and
have found that they all run at different speeds in the straight
runway. Let us suppose also that our subjects are known to differ
with respect to the amount of time that has elapsed since they
were last fed. Perhaps the fast rat has not eaten for 48 hours, the
moderate-speed rat for 24 hours, and the slow rat for 4 hours. This
type of information satisfies our need for independent data upon
which to base the definition of factor X. The deprivation histories
of the animals in this case are correlated with, but measured in-
dependently of, running speed. Different degrees of factor X can
be inferred from, or defined by, measured differences in time of
deprivation, and differences in running speed can be "explained"
by appealing to differences in factor X as thus defined. Figure 2:6
illustrates schematically the way in which several values of factor
X would be inferred from the different deprivation histories of the
three animals.

^R.

Frc. 2:6. A diagrammatic representation of the manner in which three dif-
ferent antecedent conditions. Ads, AC24, and AC4, which designate 48, 24,
and 4 hours of food deprivation, respectively, may serve as the basis for de-
fining high, medium, and low levels of factor X, in turn. Differences in factor
X defined in this way are assumed to be responsible for the fact that an iden-
tical stimulus situation (S) elicits fast, moderate, and slow rates of running
from three otherwise comparable animals.

INTERVENING VARIABLES AND DEFINITION OF DRIVE 37

In this figure, which is simply a modification of Fig. 2:3, AC48
denotes the antecedent condition of being without food for 48
hours at the time of the running-speed test, and AC24 and AC4
refer, respectively, to the conditions of being without food for
24 and for 4 hours.

2. An Intervening Variable May Be Defined in Terms of Dif-
ferences in Performance on Some Task Other than That Used in
the Test Situation. Consider again the laboratory situation in
which the responses of different subjects are found to vary though
the testing conditions are held constant, but suppose that we now
have no way of determining the past history of our subjects. Under
such circumstances it would be impossible to base our inferences
about factor X upon differences in the antecedent conditions to
which the organisms had been exposed. Such a situation is not
so likely to occur, of course, in a laboratory where the animals are
under constant supervision, but it often occurs with human sub-
jects. In most cases we simply do not know what the distinctive
antecedent events have been.

Whenever we are unable to control a subject's previous history
or to get an accurate estimate of it, we can still achieve an inde-
pendent definition of an intervening variable by measuring the
subject's responses in a second test situation. For instance, we
might take three rats and place them in separate activity cages
and record the amount of activity they exhibit during a standard
test period. Differences in amount of activity could then be used
to define levels of factor X, and these, in turn, might be invoked
to explain differences in running speed in the straight alley. This
method of obtaining an independent definition involves the formu-
lation of an R-R relationship, and is represented schematically in
Fig. 2:7.

In this diagram S2 represents the activity-cage situation, and Si
represents the alley maze. R^ R^, and R^ indicate high, moderate,
and low levels of activity, respectively. The dashed lines from R^,
R^, and R^ to the three factor-X levels indicate the process of de-
fining inequalities of this factor in terms of the responses exhibited
in the activity cages.

3. An Intervening Variable May Be Defined in Terms of Dif-
ferences in an Organic Variable. This way of attaining independ-

38 THE MOTIVATION OF BEHAVIOR

^Rr

Fig. 2:7. A paradigm to represent the method whereby differences in in-
tensity or vigor of response (R\ R^, R") in one testing situation (S2) can
yield independent definitions of the levels of an intervening variable. The
defined levels of factor X then serve to "explain" variations in response to a
different situation (Si).

ent definitions involves the measurement of some physiological
condition of the organism (O variable), such as per cent normal
body weight, blood-sugar level, stage of estral cycle, total neural
activity in the brain at some point, or skin resistance. Once such
measurements have been obtained, they can be examined to deter-
mine whether they are stable indexes of inter- or intraindividual
differences. If the three rats of our previous example were found
to differ reliably with respect to blood-sugar level, such physiologi-
cally measured values could be used as the basis for an independ-
ent definition of factor X.

Although an O variable has been described as an organic state
or characteristic of an organism, numerous instances of such states
are difficult to distinguish from responses. To take a single exam-
ple, a subject's heart rate might remain relatively constant over a
period of inactivity and hence might be termed an O variable. But
the individual contractions of the heart are clearly consequences
of, or responses to, antecedent neural impulses. Similarly, changes
in mean heart rate might appropriately be catalogued as responses.
Were these responses used to define drive differences, the proce-
dure would become an instance of the second method we have
just considered. In cases like this the distinction between the sec-
ond and third methods tends to disappear. Fortunately, this does
nothing to destroy the independence of either method.

4. An Intervening Variable May Be Defined by Referring to
Differences in Stimulus Conditions. When the characteristics of
the test situation are not the same from subject to subject, or

INTERVENING VARIABLES AND DEFINITION OF DRIVE 39

from time to time for the same subject, these differences can be
employed to obtain independent definitions of intervening vari-
ables. As an illustration, consider an experiment in which the eye-
blink response is being conditioned to a faint light {CS) by
paired presentations of the light and a puff of air {UCS) directed
at the cornea. If a strong puff is used for one group of subjects and
a weak puff for another, one can assert that the level of factor X
is a function of strength of puff, the strong-puff group being de-
fined as having a higher level of factor X than the weak-puff
group. Thus the definition of factor X is independent of the re-
sulting behavior, i.e., the frequency of conditioned eyelid responses
exhibited by the two groups.

The important elements involved in this defining procedure
are summarized in Fig. 2:8. Here the upper line denotes the con-
ditions for the strong-puff group and the lower line those for the
weak-puff group.

Thus far in this chapter we have discussed some of the reasons
why behavior theorists tend to introduce intermediary constructs
into their theories, the dangers of circular interpretations resulting
from inappropriate definitions of intermediary constructs, and wavs
of structuring definitions so as to avoid such circularity. In the

GROUP I

■w f^Ty

■'^ high

r- f^ix frequent

/
/

GROUP n

■V

■ta fT'

.-'^low

" ^"infrequent

/
/

Fig. 2:8. Elements involved in defining factor X in terms of differences in
the intensity of the unconditioned stimulus in an eyelid-conditioning experi-
ment. The greater frequency of conditioned responses (CRfrequent) exhibited
by Group I is attributed to the higher level of factor X resulting from the
use of a more intense unconditioned stimulus ( UCSg trong).

40 THE MOTIVATION OF BEHAVIOR

specific examples given, we have purposely refrained from sub-
stituting a specific label such as motivation for the colorless sym-
bol, factor X. Our next task, therefore, is to consider the problem
of how one decides whether, and under what conditions, partic-
ular names, such as habit strength, expectancy, drive, or moti-
vation, might be substituted for factor X.

Naming Intervening Variables

Although meaningful names, instead of abstract symbols, are
typically applied to intervening variables in theories of behavior,
the use of such names is not necessary, and may sometimes be
dangerous. The meaning of a factor X is given by explicit defini-
tion within the scientific vocabulary and by the nature and variety
of laws of which it is a part. Moreover, the scientific meaning of
any other term that might be substituted for factor X is given in
precisely the same ways. Thus to say that factor X is "really" drive
adds nothing to our understanding unless drive has already been
more specifically defined or has acquired more significant second-
ary meanings within the broader context of behavior theory. It is
possible, of course, that the use of a meaningful term in place
of a svmbol may facilitate communication among investigators,
but it may also lead to misunderstanding because of different pre-
conceptions as to what is meant by the "meaningful" term. It is
also likely that in the present state of psychology words must be
used because we lack so much in the way of quantification. Not
until extensive and interrelated equations and functional relations
have been established do symbols become predominant.

If one does set about providing a name for factor X, the name
must be secondary to, and contingent upon, the naming of experi-
mental variables. To make this point concrete, let us consider
again the hypothetical experiment in which three rats are found
to run at different speeds when tested in the same straight alley.
When the experiment is performed, if we have no information
about our subjects save that provided by the animals' behavior, we
can neither define factor X appropriately nor rename it. But should
we discover that the three subjects have been given different
amounts of practice in running through the alley, then practice

INTERVENING VARIABLES AND DEFINITION OF DRIVE 41

becomes a manipulated variable capable of providing an independ-
ent definition of factor X. And since practice, by convention, is
said to affect learning, factor X might then be called a learning or
habit factor. In like manner, if we were to discover that our sub-
jects differ with respect to hours of food deprivation, factor X
might be defined in terms of this variable, as in our first example
of independent definitions, and might be renamed drive. Generally
speaking, therefore, the particular name one applies to any factor
X stems from the variables used in its definition and from our
customary ways of naming the effects of those variables on per-
formance. To describe factor X as a motivational factor implies
that it has been defined in terms of certain variables, which, by
common agreement, are said to have motivational effects. But
agreements are not as common as one might wish, and we must
turn, therefore, to the task of specifying possible criteria for the
identification of motivational variables.

Criteria for the Identification of Motivational Variables

At the end of Chapter 1 it was observed that the numerous vari-
ables or conditions affecting behavior are customarily classified
into several broad groups. In elementary textbooks, chapters on
sensation and perception treat, in the main, of the effects on be-
havior of stimulus variables; chapters on instincts and develop-
ment tend to stress genetic variables; and chapters on learning,
transfer of training, and memory elevate the variables of practice
and experience to a place of primary importance. Questions often
arise as to whether certain variables belong in one or another of
these groups, but apparently fewer differences of opinion exist as
to the limits of group membership than as to the limits of the
class of motivational variables. Attempts to identify motivational
variables lead to considerable controversy, and widely accepted
criteria for making decisions in case of doubt are nonexistent. The
specific criteria listed below cannot, therefore, be regarded as either
definitive or exhaustive, or as representative of all possible views.

1 . A Variable Is Often Said to Be Motivational if It Facilitates
or Energizes a Wide Variety of Responses. This criterion, which
is probably more widely accepted than any other, stresses the fact

42 THE MOTIVATION OF BEHAVIOR

that the presence of certain variables may alter the frequency,
latency, or vigor of a number of responses. For example, a mod-
erate degree of muscular tension, produced by squeezing a hand
dynamometer, is considered motivational because it facilitates
verbal learning, mental arithmetic, the knee jerk, and a variety of
other responses (Courts, 1942). Similarly, food deprivation appears
to be motivating, especially in animals, since it often intensifies
reactions of running, sniffing, exploring, clawing, biting, whimper-
ing, and even drinking. Thus it is the nonspecific, broadly gen-
eralized effects of certain variables that seem to mark them oflF as
motivational rather than something else.

2. A Variable Is Commonly Said to Be Motivating if the Learn-
ing of New Responses Seems to Depend upon Appropriate Ma-
nipulations of That Variable. This criterion involves the assump-
tion that a reduction in the drive state associated with the variable
is rewarding and that responses followed by such a reduction will
tend to be learned. If a period of stimulation by an intense light
is terminated as soon as an albino rat presses a lever, the probabil-
ity of his making the same response on a future occasion may be
increased. The variable of intense light is then a motivator because
its offset brings about or is correlated with the acquisition of a
new response.

3. A Variable Is Sometimes Regarded as Motivational if
Changes in That Variable Lead to the Weakening of Certain Re-
sponses. Strong stimuli tending to elicit aversive or avoidant be-
havior are clear examples of variables that meet this criterion. A
child may have a strong tendency to reach out to touch a brightly
burning match, but the reaching response is likely to be aban-
doned if it is followed by a painful burn. This, of course, is the
phenomenon typically subsumed under the heading of punish-
ment, and the hot flame is treated as a motivational variable be-
cause of its efficacy in leading to the weakening of the reaching
reaction.

4. A Variable Is Occasionally Labeled Motivational Simply
Because No Other Designation Seems Suitable. If the results of
certain experimental treatments cannot be explained by appealing
to principles of learning, perception, genetics, or whatever, one
may be tempted to assert that the treatments have led to changes

INTERVENING VARIABLES AND DEFINITION OF DRIVE 43

in level of drive. For instance, altering a variable such as depriva-
tion time may result in relatively immediate and precipitous
changes in performance. If these changes are more abrupt than
we might predict from the suppositions that learning or inhibitory
processes are involved, we may wish to assume that the variable
is functioning motivationally. This criterion is perhaps the least
satisfactory of all, however, since the variable is named by exclu-
sion. By this standard a variable is motivational if it is not affect-
ing behavior as we would expect it to be affected by other
variables for which we already have meaningful names. For this
criterion to be of real utility, we would have to have much clearer
notions than we now have of what each of the many nonmotiva-
tional variables is and of how it functions as a determinant of
behavior.

It has been asserted in the previous section that, generally speak-
ing, variables are regarded as motivational if they aflfect behavior
in ways that are commonly said to be motivational. The four
criteria listed above may be regarded as a tentative summary of
some of the "motivational ways" in which behavior may be al-
tered by variables. At the present stage of our knowledge, of
course, it would be foolhardy to maintain that all students of be-
havior would approve of this list or that any one of its loosely struc-
tured criteria is entirely adequate.

Before leaving this topic we should note that certain criteria for
the identification of motivational variables are clearly not satis-
factory. For example, an increase in the probability or vigor of a
single reaction following the manipulation of a variable is not
suitable since the same behavioral effect might be ascribed, per-
haps with equal reason, to learning, expectancy, perception, or
whatever. Improvement in performance, as such, does not point
unerringly to motivation as the only responsible variable. It is also
clear that the intrinsic properties of a variable do not suffice to
distinguish it as motivational. An allegedly motivating electric
shock may be identical in its physical properties to a shock used
in studying tactual sensitivity. In like manner, one cannot rely
upon the descriptive properties of behavior alone for clear guidance
as to what is motivational. The appearance of an emotional tan-
trum in a frustrating situation may mean either heightened motiva-

44 THE MOTIVATION OF BEHAVIOR

tion or simply the transfer to that situation of a previously well-
learned mode of adjustment to frustration.

One is under no obligation, of course, in doing research, to
specify whether a variable is or is not motivational. One can, for
instance, study the effect of failure instructions upon the per-
formance of human subjects even when one is uncertain as to
how the variable should be classified. Such instructions could
operate as stimuli to elicit habits or sets; they could change the
subjects' attitudes; or they could alter motivation. The theoretical
implications of the research would perhaps be enriched if one
could decide among these alternatives, but the empirical results
would be valuable even if a decision were never made.

Operational and Significant Definitions of Drive

Operational Definitions. In an earlier section of this chapter,
mention was made of several techniques for introducing constructs
such as drive into one's theory without having to refer, in the defi-
nition itself, to the behavior to be explained by the construct. In
exemplifying these methods the definitions were never formally
presented, but each could have been stated in such a way as to
qualify as an operational definition. The need for operational defi-
nitions of concepts and terms has been heavily emphasized in recent
years by psychologists seeking a maximum degree of precision and
rigor. While the phrase operational definition may sound formida-
ble, it is essentially nothing more than a clear definition. It is a
definition in which the conditions under which the concept is to
be used have been clearly and unambiguously stated in terms of
operations or activities of observing and recording that can be
made by any competent observer. Thus the statement, "Degree of
drive in the rat is defined in terms of the time during which the
rat has been without food," is a good (operational) definition. It
states precisely what operations, i.e., measuring and recording the
time during which a rat does not eat, must be carried out to satisfy
the definition. Presumably, all observers who have stop watches or
other time-measuring devices can agree as to how much time has
elapsed since the rat has been fed. From the definition, therefore,
they can agree that rat X has a higher drive than rat Y if rat X

INTERVENING VARIABLES AND DEFINITION OF DRIVE 45

has been without food for a longer time than rat Y. When a con-
cept is defined operationally, therefore, it is defined in terms of
communicable, repeatable manipulations and observations that
can be performed by any reasonably competent observer. It is a
definition whose meaning within the scientific vocabulary is clear
because the meanings of its constituent words and symbols can be
given in terms of observable properties of physical objects and the
relations among them. Many of the definitions of drive that have
been seriously proposed in the past fail to meet the test of opera-
tional precision, but allusions to them still appear frequently in
psychological literature. Taken seriously and uncritically, nonopera-
tional definitions lead to little but confusing and interminable
arguments.

Although an operational definition of drive in terms of hours of
food deprivation may seem sensible on intuitive grounds, many
other possible definitions, though less obviously reasonable, are
equally acceptable as to their operational bases. For example, there
is no a priori reason why the strength of a man's drive cannot
be defined in terms of the length of his nose as measured under
certain carefully prescribed conditions. To assert that men with
long noses have a strong drive and those with short noses a weak
drive is a perfectly clear operational definition. It is operationally
adequate because it specifies the conditions under which the terms
strong drive and weak drive are to be used. Any person with a
suitable ruler and a little patience can presumably measure the
noses of a number of individuals, and the data he obtains will
correlate positively with those obtained by other nose measurers.
If the measurement conditions of the definition are carefully met,
all members of a subject population can, with a high degree of
agreement among observers, be separated into relatively long- and
short-nosed groups. And the observers can then agree that, by
definition, one group of subjects has a strong drive and the other
a weak drive. In a similar manner, one could formulate opera-
tionally precise definitions of drive in terms of the color of a
rat's fur, in terms of the ratio of its weight to the amount of curva-
ture of its nose, in terms of the length of steps it takes while run-
ning, or, in fact, in terms of any conceivable quantities one could
measure. As long as the measurements, discriminations, or other

46 THE MOTIVATION OF BEHAVIOR

activities that must be performed by an observer are specified
precisely by the definition, and those activities can indeed be
executed in the same way and with the same outcome by independ-
ent competent observers, the definition satisfies the requirement
that it be operational.

Significant Definitions. The mere fact that a definition is indeed
operational does not guarantee that the construct thus defined
will be generally useful or theoretically significant. Drive can be
operationally defined in terms of the length of a man's nose, and
nose length can be measured with reasonable precision. But it
would be foolish to use such a definition for very long if one could
not show that nose length meets one or more of the criteria of
motivational variables or is meaningfully related to other variables.
Thus, nose length provides a useless definition of drive unless it
can be shown that long-nosed subjects behave as though they were
more highly motivated than short-nosed subjects. Irrespective of
the elegance of our operational definitions, they are of little value
until the defined concepts have been shown to have sensible, clari-
fying relations to other concepts and to other facts. A definition
is useful or significant when the laws of which the defined concept
is a component fit meaningfully into a broader theoretical struc-
ture and serve to illuminate a variety of lower-order laws or func-
tions. Nonoperationally defined constructs can probably never be
scientifically significant; but a construct can be immaculately opera-
tional without being helpful in any way.

A distinction is thus drawn between (1) the basic scientific-
vocabulary meaning of a concept, as the result of its having been
defined in specific operational terms, and (2) the additional,
more significant meanings it may acquire if it proves to be useful
or helpful. A clear understanding of these two kinds of meanings
(Bergmann, 1944; Spence, 1948) helps to eliminate confusion in
dealing with psychological problems, and we shall have occasions
throughout the remainder of this text to refer to them again.

Since drive or any other construct can be defined operationally
in many different ways, decisions as to which definition will be
used must rest upon the utility of the concept. Of several opera-
tional definitions, the one that is most useful or significant is
clearly to be preferred. And whenever utility can be enhanced by

INTERVENING VARIABLES AND DEFINITION OF DRIVE A7

altering a definition, then such remedial measures are clearly in-
dicated. Unfortunately, we cannot state in precise language what
is meant by maximum utility or significance, and hence cannot
point to any single definition of drive as the one that is most use-
ful. The process of evaluating significance and of refining defini-
tions is an ever-continuing one, and widely satisfying answers can
probably not be expected for many years. It is even possible that
no definition of drive will ever turn out to be useful and that the
concept will disappear entirely from our scientific vocabulary.

The Problem of Drive Quantification

General Considerations. The problems involved in attempting
to quantify or to measure the strength of a drive or motive are
extremely complex, and any comprehensive discussion of them
would take us far beyond the intended scope of this book. But
the "measurement issue" is repeatedly raised in discussions of
motivation, and since it is closely allied to the definitional matters
we have just considered, it seems desirable to examine it briefly
at this point.

The bases of our concern with measurement, both in the affairs
of our daily lives and in our scientific pursuits, have been discussed
quite frequently and need only to be touched upon here. We are
constantly making judgments about the properties or qualities of
objects and situations, and our behavior is often guided by the
outcome of these judgments. A room may be judged to be "too
warm," so we remove our jackets; an automobile is seen to be
approaching the intersection "too rapidly," so we step back until
it has passed; our friends may be judged to be "too angry" at the
moment, so we may defer a request for a favor. In many instances,
judgments of this kind, in which we simply affirm or deny the
presence of qualitative characteristics, provide adequate support
for the normal activities of living. In other circumstances, however,
and especially in the pursuit of scientific knowledge, we find it
necessary to ascertain, if possible, the precise degree of the prop-
erty or characteristic about which judgments are made. Thus we
may need to know, not just that a rat has become heavier when
fed certain foods, but whether it has gained 10 grams or 50 grams;

48 THE MOTIVATION OF BEHAVIOR

we may need to know whether a subject's skin resistance has
changed by 100 or 1,000 ohms; whether one subject is more highly
motivated than another, and if so, by how much.

In the case of motivation or drive, as in many other instances,
answers to such questions as "how much," or "how intense," are
fundamental to an increase in the accuracy of our judgments about
motivation and to the discovery of comprehensive explanatory
principles. Not until these answers can be given shall we be in a
position to formulate principles or theories capable of being un-
ambiguously confirmed or refuted (Cohen and Nagel, 1934).
When we can say that an individual has a certain amount or
strength of drive we have substituted quantitative distinctions for
qualitative ones and have made an important first step toward
the measurement of the attribute or property of drive.

Acknowledged authorities on the topic of measurement, though
they differ with respect to a number of issues, hold comparable
views as to what constitutes the essence of measurement. Broadly
conceived, measurement is the process of assigning numerals to
events, objects, or the properties of objects, in accordance with
clearly specified rules and procedures. It is a process through which
experimentally demonstrable properties or relations of objects or
systems are juxtaposed against a numerical system having corre-
sponding properties and/or relationships. As Stevens (1951) has
so aptly phrased it, when we measure ". . . we deputize the nu-
merals to serve as representatives for a state of affairs in nature
. . ." (p. 23).

From these general statements about measurement, it follows,
for example, that the heights of a number of discernibly different
individuals have been measured when we have applied yardsticks
or other length-measuring devices to these individuals in certain
agreed-upon ways, and when, as a consequence of these experi-
mental operations, we have allocated one numeral to each individ-
ual. By the same token, it would be permissible to say that we
have measured the strength of a drive or motive when, by fol-
lowing certain operational procedures or rules, we have been able
to assign numerals to different organisms that are presumed to
possess different degrees of drive.

In actual practice, however, any of several different rules and

INTERVENING VARIABLES AND DEFINITION OF DRIVE 49

procedures might be followed in attempting to assign numerals to
objects or individuals possessing a certain property. As a conse-
quence, one cannot speak simply of just one kind or variety of
measurement. Typically, at least two, and sometimes four, types of
measurement are described, depending on the kinds of operations
that can be performed with the property being measured, on the
existence of related numerical laws, and on the types of mathe-
matical transformations that can be applied to the measured data.

Let us now consider several of these kinds of measurement as
they might relate to the problem of quantifying drive or motiva-
tion.

Counting. Perhaps the simplest example of what some writers
describe as a crude sort of "measurement" is that in which in-
stances of the phenomenon under study are counted or enumer-
ated. By the use of a standard set of ordered symbols, such as the
series of numerals 1, 2, 3, etc., we can, by pairing off one instance
of the phenomenon against each numeral, determine whether
there are more instances of the event in one situation or under one
set of conditions than in another. Thus we determine whether
there are more students in classroom A than in classroom B by the
simple expedient of counting the students in each room. Counting
satisfies the broad criteria of measurement because it involves
assigning the numeral 1 to a particular student, the numeral 2 to
another, and so on, until all of the students have been given a
number. If the last numeral assigned to classroom A stands higher
in the series of numerals than the one last assigned to classroom
B, we can conclude that A contains more students than B. This
procedure will not be successful, of course, unless we have some
means of unequivocally identifying the individual students so that
no two students will be assigned the same numeral and no student
fails to get a numeral.

This elementary method of measurement might be applied to
the quantification of drive in the following manner. Since count-
ing cannot be successful unless we can recognize the things or
properties to be counted, our initial step must be to set up definite
criteria for determining whether an individual is motivated. For
this purpose we might choose a simple operational definition such
as "an animal is motivated (has some degree of drive) if it is

50 THE MOTIVATION OF BEHAVIOR

awake and walking about, but not if it is asleep." Since we can
readily observe whether an animal is active and awake or asleep,
we can apply the definition to each and every animal in a group.
Thus we can recognize the presence or absence of the phe-
nomenon we wish to measure and can go on to count the num-
ber of animals in the group that are, by the terms of the defini-
tion, possessed of drive. Should we be interested, we could then
note whether colony A contains more motivated animals than
colony B, whether the count of motivated animals is higher under
one diet than another, and the like. Moreover, if it seemed ex-
pedient to do so, we could alter our initial operational definition
of drive or choose a completely different one. By means of new
counts we could then determine whether the use of the alternate
definition yields more significant or more meaningful relations
than did the original one.

The application of operational definitions and counting proce-
dures in this suggested manner does not, of course, provide us
with information as to the relative drive strengths of different ani-
mals or of different colonies. But the fact that we have been able
to divide our animals into "driven" and "nondriven" groups and
to enumerate the constituents of each group does mean that we
have achieved one kind of drive quantification.

Ranking. The operational definition employed in the previous
section, that an animal is motivated if it is awake and moving
about, is an all-or-none type of definition. By applying it we can
decide whether any one individual is motivated or not. But the
definition does not provide us with a method for determining
which of two motivated individuals is the more highly motivated.
Assertions about the relative drive strengths of different individ-
uals imply a different sort of measurement than that achieved by
counting and involve different kinds of definitions and procedures.
The outcome of such procedures is an ordinal or rank-order scale
by means of which numerals may be assigned to individuals hav-
ing varying degrees of the property or dimension of drive.

The first requirement to be met in constructing an ordinal
drive scale is to develop standardized laboratory operations for
arranging individuals in order with respect to strength of drive.
We must, that is, find empirical methods to support decisions

INTERVENING VARIABLES AND DEFINITION OF DRIVE 51

such as "A has more drive than B," "B has less drive than C,"
and so forth. But since drive is an intermediary construct, it again
becomes necessary at this point to introduce an operational defini-
tion. Perhaps we might choose a definition such as the following:
if two food-deprived rats are permitted simultaneously to run
down adjacent straight alleys for food, then the rat that first
reaches its goal has the higher drive. In principle, at least, the
procedures specified in this definition can be carried out in the
laboratory, and from the outcome it can be asserted that, with
respect to any two rats, the drive of one is either greater than,
less than, or equal to that of the other, by definition. The defini-
tion states that the rats are to be run simultaneously, and it is im-
plied that the observer simply judges which rat is the "winner."
However, the rats could be run individually, provided their com-
parative running speeds are evaluated through the use of stop
watches or other timing devices. The two methods would yield
identical estimates of relative running proficiency, and hence of
relative drive strength, unless social or other factors were differen-
tially involved in the two situations. Incidentally, other operational
definitions than the above (e.g., definitions involving differences
in antecedent conditions, organic states, or stimulus conditions)
could be used with equal justification as the basis for the rank
ordering of subjects according to drive level.

The logical requirements for quantifying any physical property
or quality have been clearly stated by such writers as Campbell
(1921), Cohen and Nagel (1934), and Reese (1943). Cohen and
Nagel have listed the minimum requirements to be met in con-
structing an ordinal scale. In the following paragraphs each of
these requirements is examined in relation to the problem of
developing an ordinal scale of drive.

Given a group of individuals, A, B, C, etc., we must be able to
arrange them serially with respect to the property (drive) so that
between any two individuals, A and B, one and only one of the
following relations holds: (a) A has more drive than B; {b) A
has less drive than B; (c) A's drive equals B's. It must also be
possible to show by physical operations that the relationship
"greater than" and its converse "less than" (symbolized by > and
< ) are asymmetrical and transitive. The requirement for asym-

52 THE MOTIVATION OF BEHAVIOR

metry is satisfied by demonstrating that if A > B, then B > A
(where > means "not greater than"). Transitivity can be demon-
strated by showing that if A > B, and B > C, then A > C.
Satisfying these requirements at the empirical laboratory level, by
comparing running times, for example, comprises the first steps to-
ward constructing an ordinal scale of drive.

The actual task of scale construction is a relatively simple matter
once we have found satisfactory physical operations for establish-
ing the fundamental relationships listed above. To do so we must
assign numerals to the individuals of the group in accordance
with certain rules. If we have determined, by laboratory tests, that
individual A has more drive than B, we must assign to A a nu-
meral standing higher in the series of numerals than the numeral
assigned to B. Conversely, if B has less drive than A, the numeral
assigned to B must be less (lower in the series) than the numeral
assigned to A. If A and B are equal with respect to drive, as ex-
perimentally determined, then the numeral assigned to one must
be the same as the numeral assigned to the other. The numeral
series is, by convention, an ordered series that exhibits transitive
and asymmetrical relations. But until we have shown empirically
that the relations among individuals with different drive strengths
are also transitive and asymmetrical we cannot profitably use the
numeral series to represent the rank orders of the individuals'
drives.

Incidentally, although our operations may tell us that A has
more drive than B there is nothing in these operations to indicate
what specific numerals should be assigned to A and B. The rule
merely states that the numeral assigned to A must be greater than
that assigned to B. Any two numerals will do: 1 and 2, 10 and 20,
or 37 and 99, so long as the second is the larger of the two. As
Reese (1943) has pointed out, at this stage we have no operations
for determining how much more drive A has than B, and hence
the numerals we assign cannot mirror a relation that has not been
determined. The ordinal scale we would obtain from assigning
numerals in compliance with these rules would tell us nothing
about the size of the interval separating any two individuals on the
scale, nor would it yield any information as to the absolute amount
of drive at any point. We can assign the numerals 2 and 1 to A

INTERVENING VARIABLES AND DEFINITION OF DRIVE 53

and B, respectively, if we wish, but we cannot maintain that A
has twice as much drive as B, since the relation "twice as much
as" has not been empirically demonstrated.

Once a rank-order ( ordinal ) scale of drive has been established,
we can use it in a number of different ways even though no mean-
ing can be attached to the statement that the drive of one individ-
ual is twice (or n times) the drive of another. For instance, any
new individual can be assigned a position on the scale with respect
to the original group used in constructing the scale. We can
also use the scale to study the effects of any new variables upon
relative scale position. Or we can relate position on the scale to
level of performance on some new and different task. Thus, we
can, from the scale, select groups of organisms with different
drive levels and subject them to a learning task to discover whether
their speed-of-learning scores stand in the same rank order as do
their drive levels. A great many of the "qualities" of individuals
that are "measured" by the psychologist, such as intelligence,
honesty, and aptitude, etc., are measured only in the limited sense
that different degrees of the quality can be arrayed in serial order.
But it is also true that certain properties of purely physical systems,
such as density and hardness, are also restricted to measurement
of this kind. All of these qualities are commonly described as
intensive, and when, for a given quality, physical operations have
been worked out which satisfy the criteria listed above, one speaks
of an intensive dimension.

In concluding this discussion, we must note that the major prob-
lem in constructing an ordinal scale of drive is to find satisfactory
empirical methods upon which to base the assertion that individ-
ual A has more drive than individual B, or that individual A has
more drive at one time than at another. The method used in our
illustration, of defining drive operationally in terms of relative
running speeds, would probably not be entirely satisfactory to
most investigators. In principle, one can construct an ordinal scale
of drive, but it will be of limited interest until a genuinely sig-
nificant definition of drive can be formulated as the basis for the
empirical process of rank-ordering subjects with respect to drive.

Extensive Properties and Fundamental Measurement. The ordi-
nal type of measurement as applied to the problem of drive quan-

54 THE MOTIVATION OF BEHAVIOR

tification can never tell us how much more drive one subject has
than another. To obtain this information it would be necessary to
demonstrate that drive intensity is an extensive dim.ension and
susceptible, therefore, to what the physicist calls fundamental meas-
urement. According to Campbell (1921), an extensive dimension
is like an intensive one in that the characteristics of asymmetry
and transitivity can be shown to hold among events having the
extensive property. The two differ, however, in that a quality that
is extensive possesses also the attribute of additivity. The prop-
erty of objects which we call their length is extensive because ex-
perimental operations have been found by which one length can
be added to another to produce more of the same property of
length. It is because length is additive, primarily, that it is said to
be extensive and to permit fundamental measurement.

At present there seems to be little reason for supposing that a
property such as drive can ever be measured in this fundamental
sense, since experimental operations for demonstrating additivity
may never be found. It is conceivable, however, that satisfactory
quantitative estimates of drive can eventually be obtained by what
is termed derived measurement. We cannot pause to describe this
method in detail here but it is applied by the physicist to dimen-
sions such as density that do not exhibit the property of additivity.

Summary

The discussions in Chapter 2 have dealt with the general ques-
tions of why intervening variables such as drive are used in be-
havior theories, how such impalpable entities can be defined by
reference to observable variables, how one might decide that a
given intermediary variable is motivational rather than something
else, and how such variables might, in principle, be measured.

In the first part of the chapter it was noted that intermediary
variables or constructs are usually proposed in the hope that they
will add to our understanding of obscure relations between be-
havior and other variables both inside and outside of the organism.
Typical relations of this kind were cited with special emphasis
being placed upon disproportions between the vigor of a response
and the intensity of a stimulus, upon behavioral variability in the

INTERVENING VARIABLES AND DEFINITION OF DRIVE 55

presence of constant environmental conditions, and on invariant
behavior in the face of normally effective changes in the environ-
ment.

Intervening variables are sometimes defined in terms of the
particular responses that are supposed to be explained by such
variables. This practice leads to circular explanatory reasoning of
an undesirable kind, as when a herding instinct, defined in terms
of gregarious behavior, is alleged to be the cause of such behavior.
To avoid such circularity an intervening variable can be defined
in terms of variations in the organism's life history, in terms of
other responses recorded in different testing situations, or in terms
of organic variables and contemporary environmental conditions.

The scientific meanings of an intervening variable are provided
by explicit definitions as to how the term shall be used in the
scientific vocabulary and by the diversity and kinds of laws into
which the variable enters. It is not necessary, therefore, that the
constructed variable be given a name that is also meaningful to
laymen. Usually, in such an attempt, the choice of a particular
name seems to depend primarily upon the independent variables
used in defining the construct and upon traditional ways of nam-
ing the effects of such variables upon performance.

Although opinions differ as to the nature of motivational vari-
ables, several criteria for their identification are suggested in
this chapter. Thus a specific variable tends to be regarded as mo-
tivational ( 1 ) if it tends to facilitate or energize several different
responses, (2) if its termination or removal following a new re-
sponse leads to the learning of that response, (3) if sudden in-
creases in the strength of the variable lead to the abandonment of
responses, and (4) if its effects on behavior cannot be attributed
to other processes such as learning, sensation, innate capacities,
and sets.

In later sections of the chapter we have pointed out that inter-
mediary constructs, including drive or motivation, must be defined
with sufficient clarity so that investigators with comparable train-
ing can understand what the terms mean. This goal can be at-
tained if construct-names are defined either in terms of directly
observable things, or in terms of words that have been so defined.
Definitions of this kind are said to be operational, and the scientific-

56 THE MOTIVATION OF BEHAVIOR

vocabulan^ meaning of the construct is thus given by the opera-
tions or manipulations used in its definition. A construct may be
clearly defined, however, and yet have no value as an element of
a theor}^ of behavior. The student of motivation, therefore, must
face the problem of how to formulate a conception of motivation
that is both operationally immaculate and generally useful. When
functioning as an integral component of a behavior theor}^,
a construct acquires, from its interrelations with different
constructs and laws, other meanings than those provided by its
operational definition. A construct can have operationally clear
meanings without these secondary meanings, but not the reverse,
save perhaps in highly developed physical theories.

Some elements of the problem of drive measurement are re-
viewed in the final section. There it is observed that measurement,
in essence, is the assignment of numbers to objects and events in
accordance with certain rules and conventions. Counting is one
way of assigning numbers, and if operations for detecting the pres-
ence of drive can be devised, then the frequency with which
"driven" subjects appear in a given population can be determined.
At a more advanced level, an ordinal scale of drive can be devised,
provided one can find physical operations for arranging individuals
in rank order with respect to drive strength. No procedures have
yet been devised by means of which drive may be measured in
the fundamental sense that length and weight are measured, but
the quantification of drive by what is termed derived measure-
ment may eventually prove to be feasible.

CHAPTER

Primary Sources of Drive

In this chapter we shall deal primarily with experimental studies
of the effects of certain alleged motivational variables upon the
performance of animals in a variety of situations. The variables to
be considered are those that are commonly said to be responsible
for the arousal of the "primary drives." For reasons presented be-
low, however, we have chosen to depart somewhat from conven-
tional terminology and to speak, not of "drives," but of "primary
sources of drive." Since these and other terms to be used hereafter
require clarification, we begin with a brief terminological discus-
sion.

Some Terminological Distinctions

Motivating and Steering Variables. With few exceptions, con-
temporary theorists make a distinction between independent vari-
ables that seem to have motivational effects upon behavior and
those that direct or guide behavior (e.g., Tolman, 1932; Lewin,
1938; Hull, 1943; Spence, 1956). Moreover, two kinds of interme-

THE MOTIVATION OF BEHAVIOR

diary constructs are introduced that correspond with these two
groups of variables. Thus, drives, motivations, conations, emotions,
and hbidos function as the activating agents; while cognitive maps,
associative tendencies, and habit strengths serve, in conjunction
with external and internal stimuli, to determine the direction be-
havior will take.

In some instances, however, difficulties arise in attempting to
maintain these distinctions. Specifically, there are certain manip-
ulatable variables that appear to exert both a motivating and a
guiding influence upon behavior and hence cannot be put into
a single classification. Peripheral shock is a good example, since
it has sensory consequences in addition to its presumed drive-
arousing effects.

Two solutions to this difficulty merit consideration. First, one
might assume that a variable such as electric shock affects only
drive, but that drive can function as both an activator and a
director. This conceptual arrangement is represented in the upper
half of Fig. 3:1. Unfortunately this solution seems unsatisfactory,
since the directive function it ascribes to drive is precisely the same
function traditionally reserved for cognitions or associative tend-
encies (lower half of Fig. 3:1). If both drive and habit are to be

Motivational

variable

(e.g. shock)

>■ Drive <C^

Environmental
variables

Sensory
' effects

Nonmotivational

variable

(e.g. practice in a

learning situation)

Associative
tendencies

Directive
function

Motivating
function

Directive
function

Fig. 3:1. The upper half of this figure represents one hypothesis as to how
a motivational variable such as electric shock might have both directive and
motivating effects upon behavior. For purposes of comparison, the directive
effects of environmental variables and of practice in a learning situation, as
mediated by associative tendencies, are included in the bottom half of the
figure.

PRIMARY SOURCES OF DRIVE

included in our theories, then the two should affect behavior in
different ways; otherwise only one construct seems to be required.
In the scheme of Fig. 3:1, habits (associative tendencies) operate
as behavior directors but not as activators, whereas drive exhibits
both properties. This interpretation thus seems to add little to the
clarity of our understanding, but several theorists (McClelland,
1951; Young, 1955; Seward, 1956; Marx, 1956) use the term
motive in essentially the way that drive has been used here.

A second solution that promises to be more useful is represented
in Fig. 3:2. Here the dual-purpose variable (shock) is assumed to
have two distinguishable consequents rather than one. It provides
sensory stimuli which, in conjunction with associative predisposi-
tions, can impart direction to behavior, and in addition it affects
drive, to which only the single function of motivation is assigned.
This answer has the advantage of preserving unique functions for
drive as well as for habit strength. On this view, which is tenta-
tively adopted throughout the remainder of this book, the con-
struct denoted by the words drive or motivation — these terms are
used as synonyms here and elsewhere — is assumed to have no
function as a behavior guide or director. Certain variables may
thus be seen to have dual consequences, but drive is assumed to
have only motivational effects and habit strength only directive
functions.

Motivational

variable -
(e.g. shock)

->- Drive

Motivating
function

Environmental
variables

Nonmotivational

variable
(e.g. practice in a"
learning situation)

^ Sensory
effects

Associative

tendencies

Directive
function

Fig. 3:2. Components of a second interpretation of the dual behavioral ef-
fects of a motivational variable. Here electric shock is assumed to affect drive,
which serves as the sole motivating agent, and also to lead to sensory effects.
These effects are essentially identical with those arising from other aspects
of the physical environment and are presumed to function in conjunction
with associative predispositions to direct behavior.

60 THE MOTIVATION OF BEHAVIOR

It should be clearly understood, however, that an adequate
theory of directed behavior may not require both an activating and
a steering agency. As we shall see in the next chapter, some the-
orists feel that a construct having an activating function can be
dispensed with entirely, the assorted effects of the so-called motiva-
tional variables being explained by appeal simply to changes in
stimuli and in associative strengths.

Drives or Drive? In current discussions of motivation it is com-
monplace to encounter the word "drives." For certain writers, this
term apparently conveys the idea of multiple directedness. The
hunger drive is said to be directed or to direct behavior toward
food, the thirst drive toward water, and so on. But this terminology
is confusing if, as we have argued, it is desirable to limit the
function of a drive to that of an activator or motivator. If this
latter position is adopted, drive can never be directed toward any
specific goal, nor can it selectively activate one type of associative
tendency to the exclusion of others, since this would indirectly
involve a directive function. To speak of "drives" implies that the
constructs so designated are alike, yet different. If they are exactly
alike when functioning as motivators, then identical processes
must be involved in all cases, and all drives, as activators, become
one. If they are not alike as motivators, then each must be motivat-
ing but in a unique way. Just what these different yet comparable
ways might be is difficult for one to imagine. One might sup-
pose, of course, that drives are all alike save that each is the result
of its own distinctive motivational variable. But if this is the case,
then we no longer have different drives, as behavior determinants,
but only different sources of drive. It is this line of reasoning that
has led us to entitle this chapter "Primary Sources of Drive" rather
than "Primary Drives."

The conclusions reached above, that a sharp distinction should
be drawn, whenever possible, between the driving and steering
determinants of action, and that drive, as an intermediary con-
struct, should be unitary rather than multiple are clearly coordinate
with a theory proposed by Hull (1943) in his Principles of Be-
havior. According to Hull, behavior is determined, in large meas-
ure, by two intermediaries, drive (D) and habit strength (H).
Drive, for him, is a broadly acting, nondirective factor that func-

PRIMARY SOURCES OF DRIVE 61

tions exclusively to facilitate associative tendencies whether learned
or unlearned. Drive results from the manipulation of certain vari-
ables, such as strong stimuli and the withholding of food or water.
Furthermore, its capacity to facilitate all behavior is assumed to be
independent of the particular antecedent condition of which it
may, at the moment, be a function. For example, drive due to in-
tense stimulation is indistinguishable, as an energizer, from drive
due to food deprivation. At the theoretical level, drive is simply
a numerical quantity multiplying associative quantities (habit
strengths) to yield a further quantity, excitatory potential (E).
This latter quantity, further altered by other factors, is in turn re-
lated by one or another postulated mathematical law to overt
behavior.

The behavior-directing function is performed within Hull's sys-
tem by the hypothetical associative tendencies, whether learned
or instinctive, functioning in combination with both internal and
external stimuli. An animal in a discrimination situation is steered
or directed toward one stimulus object and away from another by
its learned associative predispositions. But drive, being nondirec-
tive, is presumed to facilitate both movements of approach to
the positive cue and movements away from the negative cue.
Whichever reactive tendency is dominant at the moment is
catalyzed into overt action by drive.

Primary and Secondary Sources of Drive. Although mention has
been made of primary sources of drive, we have neither explained
the meaning of the adjective "primary" nor contrasted such sources
with those that are called "secondary." Broadly speaking, primary
motivational variables are those that produce their effects through
the action of inherited bodily mechanisms. When environmental
conditions are altered in any of a variety of ways, the physiological
mechanisms of an organism, even in the absence of opportunities
to learn, tend to react in a corrective manner. At such times the
organism is likely to behave as though motivated. The environ-
mental changes or variables leading to these effects are called pri-
mary because they appear early in the developmental and phylo-
genetic sequences, not because they are necessarily more important
than those labeled secondary. Such terms as homeostatic drives,
biogenic drives, and physiogenic drives have been used as synonyms

62 THE MOTIVATION OF BEHAVIOR

for what we are calling primary sources of drive. These terms serve
further to emphasize the regulatory mechanisms involved and
their relations to the genetic constitution of the organism. Ex-
amples of variables typically described as having primary motiva-
tional effects are the withholding of such commodities as food,
water, or air for breathing. The removal of a mother's offspring —
in some species only — is also described as a primary variable, as
are deviations from optimal levels of environmental temperature
and humidity, and pressures produced by the accumulation of
bodily wastes. Any stimulus to which the organism is receptive
may have motivational effects if it is intense enough. Electric
shocks, bright lights, loud noises, pin pricks, and the like, are com-
mon instances. In the opinion of some authors, the withholding
of opportunities to play, to be active, or to explore, also qualify as
primary motivational variables.

The secondary drives, or, as we shall describe them, the second-
'ary sources of drive, differ from their primary counterparts in that
their efficacy as motivators rests largely upon learning. Specifically,
if an individual has acquired a tendency to make a certain response
to a particular environmental situation, the elicitation of that re-
sponse may have motivational consequences; consequences, that is,
resembling those stemming from primary sources of drive and con-
sistent with one or more motivational criteria. Because human be-
havior is often said to be largely motivated by secondary or ac-
quired sources of drive, these have come to occupy the attention
of psychologists to an increasing degree. All of Chapter 5 is
devoted to their analysis.

The remainder of the present chapter is concerned with a dis-
cussion of certain primary sources of drive and their effects upon
behavior. Since hunger and thirst are the most frequently studied
sources and since their strength is often believed to be reflected
rather directly in the amount of consummatory behavior, we
begin with the question of how such behavior is regulated.

The Regulation of Consummatory Behavior

Students concerned with the ways in which biological needs may
serve as primary sources of -drive have dealt extensively with the

PRIMARY SOURCES OF DRIVE 63

effects of such needs on performance in problem solving, discrimi-
nation, and conditioning tasks. In addition, considerable effort
has been devoted to the study of the physiological mechanisms
controlling consummatory behavior. Most of the relevant research
in this latter area lies more properly within the field of physiology
rather than in that of psychology; hence we shall consider it only
briefly and in broad outline. For more detailed information the
reader may wish to consult such sources as Morgan and Stellar
(1950), Stellar (1954), Miner (1955), and Morgan (1957).

The Local Theory of Hunger and Thirst. When psychologists
and physiologists first became concerned with the nature of hunger
and thirst, they tended in large part to approach the problems
from the basic position of the introspectionists. Looked at in this
way, hunger and thirst were sensations experienced by the self-
observing scientist, and the proper way to study therh was to seek
for meaningful relations between these sensations and other con-
ditions such as time since eating or drinking and organic activities.^
Such an approach to the study of these needs is well illustrated by
the work of the physiologist Cannon (1929). Over a period of
many years. Cannon's research was directed toward discovering the
physiological correlates of these sensations and toward elaborating
and justifying his "local theory" of their origin. The essence of
this theory was that thirst, or at least the sensation of thirst, was
a consequence of a dryness of the mouth and throat caused by th^
body's need for water. Similarly, hunger was identified with sen-
sory impulses arising from vigorous contractions of the empty
stomach. The label "local theory" has been applied to this view
because of its emphasis upon the peripheral, localized origins of
the hunger and thirst sensations.

The evidence presented by Cannon and others to support his
views was extensive and persuasive. If dryness of the buccal cavity
is alleviated by rinsing the mouth with water, the sensation of
thirst is also reduced, at least temporarily, even though no water
passes into the body. And any one of a wide variety of events, such
as tightening one's belt, smoking, or becoming frightened, may
temporarily reduce or eliminate hunger contractions and thereby
also alleviate the subjective pangs of intense hunger. Moreover,
objective records of gastric contractions obtained from subjects

64 THE MOTIVATION OF BEHAVIOR

who swallowed special recording devices (Cannon, 1929) were
found to coincide with subjective reports of hunger pangs. Hunger
also appeared to be related to level of overt activity, since the
movements of sleeping subjects occurred at about the time of
vigorous stomach activity.

The local theory of hunger and thirst thus seemed adequate
to explain the origins of the sensations consequent to food and
water deprivation. But with the growth of interest in the broader
consequences of deprivation, especially in animals, it soon became
clear that the local theory could not encompass all of the impor-
tant motivational phenomena of hunger and thirst. Its inade-
quacy was due, in part, to the fact that introspective methods
could not be used with animals, but more importantly to the find-
ing that laws obtained when hunger and thirst were defined as
sensations were different from, and apparently not as useful as,
those obtained when hunger and thirst were defined in terms of
consummatory behavior. For example, thirsty subjects do not cease
drinking immediately after the first mouthful, even though only
a small quantity of water is needed to moisten the mucous mem-
branes of the mouth and throat and thus eliminate the sensations
of thirst. Nor do subjects stop eating after their first few bites, in
spite of the fact that their hunger contractions have doubtless
ceased and their hunger sensations have been allayed. Hunger and
thirst as sensations thus turned out to be different from hunger
and thirst as regulators or motivators of eating and drinking.

Physiological Mechanisms Governing Consummatory Behavior.
Amount or rate of consummatory activity has been used as the
principal basis for estimating biological needs in a great many
physiologically oriented investigations of basic mechanisms. A de-
tailed review of factors responsible for the control of consumma-
tory activity would take us far beyond the intended scope of this
book (see, for example. Miner, 1955) but a brief summary of some
of the major trends appears justified.

Concerning the factors that function to produce cessation of
eating, Grossman (1955) has concluded that they may be divided,
for purposes of analysis, into four components: (1) oropharyngeal
factors, (2) gastrointestinal factors, (3) circulating nutrients in
the blood, and (4) stored nutrients in the tissues.

PRIMARY SOURCES OF DRIVE 65

The oropharyngeal component refers to the stimulation of sen-
sory receptors in the mouth and pharynx produced by food in the
mouth and by the subsequent movements of chewing and swal-
lowing. Available evidence points to the conclusion that stimula-
tion of these head receptors tends to produce cessation of eating,
apparently through the mediating action of the medial portion of
the hypothalamus. The neural impulses function, it would seem,
as a kind of signal that, in cooperation with many other kinds of
signals, tells the brain to shut off the mechanisms that initiate and
maintain consummatory activity. The oropharyngeal control func-
tion, Grossman points out, is most effective when supplemented
by the entry of food into the stomach. His studies indicate that
neither factor operating alone will produce satiety. For example,
if a portion of a dog's daily ration is placed directly into its
stomach through a fistula, just before food is given ad libitum,
voluntary intake is reduced, but the dog will still eat. This it would
not do, clearly, if stomach distention provided sufficient inhibitory
stimulation. However, the quantity of food ingested during the
subsequent ad-lib feeding period is greater than if the pre-ad-lib
portion is eaten in the normal manner. Thus the suppression of
oral intake is greater when the head receptors are stimulated dur-
ing the normal processes of eating and, in addition, the stomach
is distended. Experimental studies by Berkun, Kessen, and Miller
(1952) and by Miller, Sampliner, and Woodrow (1957) have also
shown that consummatory behavior is reduced more when a given
amount of food or water is ingested orally than when it is slowly
injected directly into the stomach. When milk is very rapidly in-
jected, however, the opposite effect is obtained (Smith, Pool, and
Weinberg, 1959), suggesting that rate of injection may be a sig-
nificant variable.

In some studies, involving the use of esophageal fistulas, food
consumed orally is not permitted to reach the stomach. Animals
that are given sham feeding in this manner generally eat more
than their normal daily rations before stopping. This increase in
intake, when the oropharyngeal cues are functioning in isolation,
points to the need for supplementary control by gastrointestinal
factors; but the fact that eating ceases even temporarily when
stomach distention does not occur suggests that hunger drive, re-

66 THE MOTIVATION OF BEHAVIOR

garded as a determinant of eating, has suffered abatement. Gross-
man summarizes the situation as follows: "From these observa-
tions, we may hypothesize that stimulation of head receptors by
smelling, tasting, chewing, and swallowing, during eating, plays
an important role in bringing about satiety and suppression of
further eating, but that this factor is relatively ineffective when
it is not associated with entry of food into the stomach" (1955, p.
86). On the basis of these and other studies it appears probable
that hunger, defined in terms of eating behavior, can be controlled
by appropriate stimulation and its resulting neural activity.

Apparently, knowledge as to the precise neural mechanisms
through which stimulation of the head receptors can reduce the
tendency to eat is not yet available. Experimental evidence (Anand
and Brobeck, 1951) is at hand, however, to support the view
that stimulation of the hypothalamus may lead either to hy-
perphagia (overeating) or to aphagia (undereating). Brobeck
(1955), in reviewing the evidence for neural regulation of hunger
and appetite, has hypothesized that the lateral hypothalamus may
be the facilitative and the medial hypothalamus the inhibitory
mechanism. On this view, when food is eaten, certain changes
occur within the body which either directly or indirectly affect the
hypothalamus. "These changes serve as signals to the brain, tend-
ing to suppress the activity of the lateral hypothalamus and thus
to decrease appetite, while they stimulate the medial or inhibitory
portion of the mechanism and thus promote satiety" (Brobeck,
1955, p. 48). As the food becomes absorbed into the body, the
situation tends to reverse; the lateral hypothalamus becomes more
active and the medial portion more inhibited. Under these con-
ditions the animal's locomotor activity increases, and eventually
eating begins again if food is available.

Concerning the role played by nutrients in the blood and in
the tissues in regulating consummatory activity, we need only
note that physiologists have not yet agreed on the extent to which
either of these factors governs consummatory activity. Moreover,
the mechanisms of the alleged effects remain obscure (cf. Gross-
man, 1955).

PRIMARY SOURCES OF DRIVE 67

Needs and Drive Distinguished

An interesting problem, arising in part out of these studies of
the regulation of consummatory behavior, is that of the relation
of physiological needs to drive. The term need, as it is usually
used, refers to a bodily imbalance or departure from normality
produced by any one of a variety of conditions. If food is with-
held for a sufficient period of time, the chemical structure of the
body is altered, and a need for food is said to exist because the
ingestion of food is the necessary condition for the restoration of
the original state. Other variables such as water deprivation,
hormonal imbalances, extreme temperatures, and even noxious
stimuli can be thought of as generating needs for conditions
conducive to the restoration of equilibrium. When vari-
ables such as these are manipulated, behavior is often affected in
ways that are consistent with one or another motivational crite-
rion, and it is common, therefore, to identify these needs with
drive. Moreover, it is often implicitly assumed that drive fluctuates
concomitantly with needs as the latter are modified by the manip-
ulation of appropriate variables.

Over restricted ranges of variation and for a limited number of
needs, the identification of need with drive can perhaps be de-
fended. But as many bits of evidence show, the two concepts
must be distinguished under certain conditions. Consider, for
example, the case of the rat from which all food is withheld until
death occurs. If this animal's need for food is defined by reference
to loss of body weight, then its need increases progressively up to
the moment of death. But if, as is sometimes done, drive is inde-
pendently defined in terms of the number of times the animal
crosses an electrified grid to reach food, drive increases at first and
then decreases. Decreases in number of grid crossings with pro-
longed starvation are usually ascribed, of course, to muscular weak-
ness attending inanition. Nevertheless, the two quantities, need
and drive, when defined in these two different ways, are not
covariant save perhaps for the first two or three days of the depriva-
tion period.

It is also obvious that the body may, as a consequence of im-
proper diet, develop a physiological need for a specific chemical

68 THE MOTIVATION OF BEHAVIOR

substance such as a vitamin. But an organism with a deficiency of
this sort may not exhibit increased activity or other behavioral
characteristics typical of increased level of drive (behavioral defini-
tion). The body's need for the vitamin may increase as a function
of time, but drive, as estimated from changes in overt behavior,
remains constant.

Other evidence to support the assertion that need and drive
are not always covariant and hence should be differentiated comes
from a variety of observations. It is generally accepted, for exam-
ple, that the need for sexual gratification depends primarily upon
the presence of hormones in the blood (Morgan, 1957). The satis-
factory completion of the sex act, however, does not produce a
direct or immediate reduction in hormonal concentration. Never-
theless, strength of sex drive, conceived as a general motivating
tendency, is clearly reduced by copulation.

Studies of sham drinking and of sham eating such as those of
Adolph (1943) and Janowitz and Grossman (1949) also point to
the noncorrespondence of need and drive. Here the important
fact is that an esophageal animal will temporarily stop drinking
or eating even though no water or food is permitted to enter its
stomach. For a brief period following sham ingestion, the ani-
mal's thirst or hunger drive, defined in terms of its willingness to
engage in the consummatory act, is nonexistent. Yet the body's
need, as measured by loss of weight, tissue conditions, or hours
of deprivation, does not decline, and may even increase. Ap-
parently, consummatory activity itself may be drive reducing even
though need is not thereby altered.

These examples should suffice to support our contention that a
need, as estimated from one set of observations, may vary inde-
pendently of drive, when the latter is defined by appeal to a differ-
ent set of data. Similar lacks of covariation may be found, of
course, between two definitions of a need or two definitions of a
drive, whenever the members of the pair are defined by reference
to conditions of observation that are quite different. Thus tissue-
condition-defined need does not correspond perfectly to drinking-
behavior-defined need; and deprivation-estimated drive may not
covary with activity-wheel estimates of drive. This failure of dif-
ferent definitions to coincide is one of the persisting problems

PRIMARY SOURCES OF DRIVE 69

facing the student of motivation. Its solution, as we have aheady
suggested in our chapter on the definition and measurement of
drive, probably hinges upon the formulation of definitions that
are useful and significant as w^ell as operational.

We conclude this section with the observation that a tendency
indiscriminately to identify need with drive is frequent among
those who seek to disprove the hypothesis that drive reduction
is a necessary condition for learning. Arguments against the drive-
reduction concept are sometimes based on evidence of the follow-
ing sort: Rats rewarded with a nonnutrient solution of saccharin
and water will learn a simple instrumental response more readily
than rats rewarded with plain water. Empirically, therefore, the
sweet-tasting solution is more reinforcing than the plain water
in spite of the fact that the former is no more effective in reducing
the body's need for water and has no food value. From this it is
reasoned, assuming that the water needs* of both groups have
been equally satisfied and that their drives have therefore been
equally reduced, that drive reduction is unnecessary for the growth
of learned associations.

The suggestion that needs and drive should be differentiated
provides one plausible answer to this type of argument. Thus
saccharin-rewarded rats and water-rewarded rats might differ with
respect to drive reduction, even though their need reduction was
the same. The sweet-tasting saccharin solution would be expected
to provide a greater amount or intensity of oropharyngeal stimula-
tion than would plain water. And this, in terms of the neural
inhibitory mechanisms described by Brobeck, could readily be
followed by decreased appetite and increased satiety. Whether
this is the drive-reduction mechanism of reinforcement we cannot
say. But this analysis suggests that at least one kind of drive-
reduction mechanism could be excited by saccharin even though
need for water is not differentially affected. A recent observation
by Smith and Capretta (1956), though nonsystematic, supports
this interpretation. According to these authors, after rats have been
injected with sufficient insulin to produce insulin shock, they are
less likely to manifest severe shock symptoms if they are allowed
to consume saccharin. Apparently the neural consequences of the
sweet-tasting substance set into motion the same kind of com-

70 THE MOTIVATION OF BEHAVIOR

pensatory machinery normally excited only by the ingestion of
sugar.

Along this same line of thought, the mere sight or smell of food
(stimuli usually described as secondary reinforcers) may be effec-
tive as rewards for new learning simply because they do lead to a
reduction in drive. Because of the individual's repeated oppor-
tunities to see and smell food Just prior to its ingestion, associa-
tions could be formed having the capacity to produce a temporary
and partial decline of drive. Perhaps the cues provided by food
come to evoke a competing reaction that interferes with events
in the hypothalamus normally responsible for hunger-drive be-
havior. Direct experimental evidence to support this notion is
lacking, but several writers (Mowrer, 1951; Osgood, 1953; Farber,
1954cz; Brown, 1955) have hypothesized that such a relationship
might exist. Even Tolman (1949), who seldom advocated the
drive-reduction view under any conditions, maintained that sub-
goals antedating the final goal might, to a degree, be drive reduc-
ing. Moreover, Morgan (1957), who for reasons similar to those
given above also maintains that drive and need should be distin-
guished, contends that a strong argument can be presented for the
view that sensory stimulation can reduce drive. Neurophysiological
evidence to support this position is also available and is discussed
briefly in Chapter 9.

Performance as a Function of Variations
in Primary Sources of Drive

Some students of behavior, as we noted earlier, are not vitally
interested in the details of the physiological processes resulting
from manipulations of primary sources of drive. Instead they are
concerned with the observable effects of such manipulations on
the behavior of their subjects in a variety of situations. Such
research workers seek to determine the empirical laws relating
the dependent behavior variables to independent motivational
variables. Although few of these laws are as firmly established as
might be desired, a considerable number of experiments have been
directed toward their clarification and understanding. It is to the
presentation and discussion of selected examples of these types

PRIMARY SOURCES OF DRIVE 71

of experiments that the remainder of this chapter is devoted.

The Effect of Deprivation Time on Consummatory Behavior.
One of the basic research problems in the field of motivation is
that of determining the relation between time of deprivation and
consummatory behavior. For example, if an animal is deprived of
food for varying periods of time before being allowed to eat ad
libitum, what is the function relating duration of deprivation to
rate of eating or some other index of voracity? Though such a
problem is obvious and is perhaps logically an antecedent to other
problems, few experiments have been directed toward its solution,
and the precise nature of the laws for different species, different
kinds of deprivation, and the like, remain to be determined in
future experiments.

From available experiments, two may be cited as illustrations.
The first, by Siegel (1947), seems to be among the earliest sys-
tematic studies of water consumption as a function of hours of
privation. In his experiment, 60 male albino rats were used as
subjects. They were housed in a thermostatically regulated environ-
ment marked by small temperature and humidity changes. On the
first evening all animals were weighed and their intake of water
was measured over a five-minute period. Since they had been
maintained on an ad-lib drinking and eating regimen since wean-
ing, the values obtained from these first measurements provided
estimates of water ingestion following zero hours of water depriva-
tion. The body-weight data were then used as the basis for dividing
the animals into four groups of approximately equal weight. On
the second and third days the animals were given an additional
test to see how much water they would drink in five minutes.
At the time of these tests, the four groups had been deprived of
water for different lengths of time. These times were 6, 12, 24,
and 48 hours. The results of these tests are summarized in Fig.
3:3, where mean water intake in milliliters (cubic centimeters)
has been plotted against hours of deprivation.

It is clear from this figure that rate of drinking is for the most
part an increasing, negatively accelerated function of time of
privation for periods up to 48 hours. The initial portion of the
curve exhibits a slight positive acceleration, but, as Siegel observes,
this may well be due to chance. The fact that rate of drinking is

THE MOTIVATION OF BEHAVIOR

12 24

Number of hours of water privation

Fig. 3:3. Amount of water (in milliliters) drunk during 5 minutes by rats
as a function of the number of hours since drinking was last permitted. (From
Siegel, 1947.)

higher at the 48-hour than at the 24-hour point is of interest
because earher studies by Warden (1931), using the Columbia
obstruction box, had implied that thirst decreased after 24 hours.
There are various plausible reasons for the poorer performance
of Warner's 48-hour animals, such as inanition or increased sen-
sitivity to electric shock. Nevertheless, the fact remains that we
have here another instance of noncorrespondence between two
measures or definitions of drive.

A subsidiary finding of Siegel's also deserves mention. Just be-
fore testing his animals on the second and third days, he weighed
them carefully to find out how much weight the groups had lost
during their respective deprivation periods. From these data and
from measures of drinking he was able to plot per cent of body
weight drunk in five minutes against percentage of body-
weight loss. This is shown in Fig. 3:4. In its general form this
curve resembles one previously reported by Adolph (1941) relat-
ing sham drinking to body-weight deficit in dogs. It provides
further confirmation of the observation that animals tend to
adjust their intake of water to match the body's need for fluid.
And incidentally the curve provides an excellent example of what
was described in Chapter 1 as an OR law, since body-weight loss
reflects an organic state (O) and drinking is clearly a response (R).

PRIMARY SOURCES OF DRIVE

-15 -10 -5

Percentage body weight loss

Fig. 3:4. A functional relation between an organic variable (loss in body
weight) and a response variable (amount of water ingested). {From Siegel,
1947.)

Our second experiment illustrating the effects of privation on
consummatory behavior is one by Horenstein (1951 ) . In her study,
the subjects were 10 female rats that had been used in a previous
investigation of the relation of drive to response latency. At the
outset, the 23.5-hour feeding rhythm on which they had previously
been maintained was reestablished. Then the animals were satiated
on wet mash and returned to their home cages for the duration
of the deprivation interval. At the end of that interval, the sub-
jects were weighed, permitted to eat freely of mash for 20 minutes,
and were finally weighed again. The quantity of food consumed
was computed from the difference between these initial and final
weights. The deprivation intervals were: 0, 0.5, 1, 2, 6, 12, and 23.5
hours. All of the animals were tested under each of the various
deprivation conditions in a random order to balance out the
influence of systematic factors such as learning, aging, etc. The
results of these manipulations are summarized in Fig. 3:5.

As the curve of Fig. 3 : 5 clearly shows, mean food intake during
the 20-minute test period was an increasing monotonic function
of hours of deprivation. In contrast to Siegel's water-drinking data,
the food ingestion curve of Horenstein's exhibits a rapid initial
rise from zero to two hours of deprivation. This effect is apparently

THE MOTIVATION OF BEHAVIOR

^13

•| 20

^^'^

^^»'^^

'Z, 15

.^'''^^^

^^"^^

^^x*"'^

^.X"*^

i^io

f""^

_*c

O „

•^ 5

- 1

1 1 1 1 1 1

0 1 2

6 12

Number of hours of food deprivation

23.5

Fig. 3:5. Amount of food ingested during a 20-niinute period by rats de-
prived of food for varying lengths of time. {Yrom Horenstein, 19 SI.)

not specific to consummatory behavior since Horenstein obtained
a similar result in her studies of the effects of deprivation time
upon latency and resistance to extinction of a simple instrumental
response. Moreover, the phenomenon had previously been reported
by Koch and Daniel (1945) and by Saltzman and Koch (1948).
Several possible reasons for this sharp increase in response strength
during the first two hours are discussed by Horenstein, among
them being cyclic factors and digestive mechanisms of the rat.

Concerning these consummatory-response studies of Siegel and
Horenstein, two explanatory comments are indicated. First, no
vigorous instrumental responses such as crossing a charged grid
or pushing a bar, over and beyond those necessary for eating and
drinking, were required of the animals. Since the performance of
such nonconsummatory responses involves additional energy out-
put, estimates of hunger or thirst requiring their performance
might yield spuriously low values, especially at longer deprivation
times, when the animal may be somewhat debilitated. Moreover,
in Horenstein's experiment, the animals were allowed to eat wet
mash instead of hard pellets, a substitution which would minimize
the effort involved even in eating. Consequently the animals' per-
formance level was probably not appreciably reduced by inanition.

PRIMARY SOURCES OF DRIVE 75

The second comment on these studies has to do with their
potential utihty as estimates of drive (D) in Hull's sense of the
term. If drive, as Hull assumed, serves to energize into overt action
all reactive tendencies, whether learned or unlearned, then eating
and drinking curves reflect the strength of D in a relatively direct
manner, provided that the strengths of the habits to eat and drink
are equal at the different deprivation intervals. Stimulus similarity
is one of the variables believed to determine habit strength, but
in order to explain how this variable might have affected Horen-
stein's results, we must digress briefly to discuss deprivation-induced
stimuli.

Within Hull's system, deprivation, like electric shock (cf. Fig.
3:2), is assumed to generate not only D, the nondirective catalyst,
but also, in most cases, distinctive internal stimuli. These stimuli
are consequences, for example, of stomach contractions attending
hunger and of dryness of the throat during thirst. These internal
events have sensory components and, unlike D, can have a direc-
tive influence on behavior since associative connections can be
formed between them and overt (as well as implicit) reactions.
Hull used the term drive stimuli for these internal cues and sym-
bolized them by the notation So- This designation is somewhat
misleading, however, since the internal stimuli neither originate
in nor bear any direct relation to drive (D) as such. It would be
more appropriate, though less succinct, to call them motivation-
variable stimuli (Smvs), since they, like drive, result from manipu-
lations of so-called motivating variables.

Now in Horenstein's experiment, though not in Siegel's, the
animals had been maintained for a good many days on a fixed
dietary regimen, being fed every 23.5 hours. Because of this regu-
larity, as Horenstein suggests, the tendency to eat may have
become more strongly associated with internal stimuli characteristic
of a 23.5-hour deprivation period than with cues arising from
shorter privation times. Thus the fact that Horenstein's curve was
highest at the 23.5-hour point and lower at increasingly shorter
deprivation intervals might be due, in part, to differences in asso-
ciative strengths. This would follow if as the privation periods
became shorter and shorter, the internal cues became increasingly
different from those of the 23.5-hour training interval, and if.

76 THE MOTIVATION OF BEHAVIOR

with increasing disparity, there was a decrease in the associative
tendency to eat. This sort of nonmotivational interpretation of
data deserves serious consideration, and in the next chapter we
shall discuss the matter at greater length. The important conclu-
sion to be reached here, however, is that changes in performance
following the manipulation of a motivational variable cannot be
ascribed solely to changes in drive unless there is reason to believe
that associative strengths have remained constant. In this respect
Siegel's results come closer to providing a "pure" estimate of
drive, since his animals were not habituated to an externally im-
posed drinking schedule prior to the experiment, and the drinking
responses, therefore, could hardly have become associated with
only one specific value of thirst-stimulus intensity. The ad-lib
drinking schedule under which Siegel's rats were reared might
have led, however, to the growth of associations between drinking
responses and various low-intensity thirst stimuli. The presence
of such tendencies would tend to encourage drinking at short-term
privation test periods relative to long-term periods.

Estimates of drive strength based upon consummatory behavior
may also be affected by the number of previous experiences the
organism has had with deprivation and with the removal of
deficits by eating or drinking. As Ghent (1957) has clearly shown,
when rats are permitted to eat or drink after their first deprivation
experiences, they do not immediately go to a familiar food dish or
to a familiar water spout. But with repeated privations, eating and
drinking latencies decrease markedly, and time spent in consuming
food and water increases. Corroborative evidence to support this
view that prior experience with deprivation affects consummatory
responses has been reported by Young (1949), Baker (1955), and
Lawrence and Mason (1955).

Instrumental Behavior as a Function of Deprivation. The two
experiments we have just examined were chosen to illustrate the
effects of deprivation upon behavior which, though probably
learned in part, appears quite early in the organism's life and
under conditions seldom controlled by the experimenter. In this
section we shall review experiments in which the indicant reactions
are instrumental to the obtaining of food or water and have been
learned under the supervision of the experimenter. Thus we shall

PRIMARY SOURCES OF DRIVE 77

be dealing with the effects of motivational variables upon such
responses as running, swimming, and lever pressing, which must
be performed by the subject to obtain a reward. As in the pre-
ceding section, we shall make no attempt to review existing
studies in toto, but will concentrate on a few of the more recent
and more adequately controlled experiments.

One problem of central importance in this area is that of the
effects of different levels of a motivational variable upon the rate
of acquisition of an instrumental response and upon terminal
levels of proficiency. In principle, the experimental design required
for this problem is extremely simple: different groups of subjects
are taught the to-be-learned reaction under different levels of the
motivational variable, and their performance during acquisition
is observed and recorded. Unfortunately, our psychological litera-
ture contains few studies of this simple, straightforward type.

An experiment by Fredenburg (1956), designed in essentially
this way, yields evidence as to the effects of different deprivation
levels upon the performance of a simple instrumental response. In
her study two groups of rats were trained to run down a 4-foot
alley for food. One group received this training after 3 hours of
food deprivation, the other after 22 hours of deprivation. Two
additional groups, also under 3- and 22-hour deprivation condi-
tions, respectively, were trained in an II -foot alley. Four trials
per day were given to each rat for a period of 12 days. By means
of photoelectric devices it was possible to measure both starting
and running time on every trial. Starting time was defined as the
interval between the opening of the starting-box door and the
instant the rat had moved a distance of I foot out into the alley.
Running time was the time taken to traverse the second, foot-
long segment of the alley.

Acquisition curves for the two groups trained in the 4-foot alley
are reproduced in Figs. 3:6 and 3:7. These speed curves were
obtained from the starting and running times by converting each
individual subject's time scores into reciprocals (by dividing each
score into 1) and then averaging these reciprocals. As the curves
in these figures show, the performance of the 22-hour deprivation
group was superior to that of the 3-hour group in both starting
and running speed. Comparable results favoring a long- over a

78 THE MOTIVATION OF BEHAVIOR

•e 2

22 hours privation
(JV=12)

3 hours privation
(N=12)

3 4 5 6 7 8 9
Successive four -trial blocks

10 11 12

Fig. 3:6. Mean starting speeds over successive blocks of four trials of rats
that had been deprived of food for 22 and for 3 hours. {Adapted from
Fredenburg, 19 S 6.)

short-deprivation group were also obtained in the case of additional
animals that Fredenburg trained in the 11-foot alley.

The conclusion to which these data point is amply supported
by other experimental evidence. For example, Loess (1952) found
performance level during learning, as measured by speed of re-
sponse, to be higher for a 22-hour group than for a 3-hour group.
In his experiment the response was one of running from a starting
compartment into a choice compartment to touch a small bar
projecting from the back wall. Kimble (1951), Ramond (1954),
Davenport (1956), and Bass (1958) have also shown that acquisi-
tion curves obtained after long deprivation intervals tend to reach
higher final asymptotic levels than curves obtained following brief
periods of deprivation. Moreover, performance of an instrumental
running response is better under high than under low drive even
when these conditions are induced in the same subjects on alter-
nate days (Bass). We may take it as reasonably well established,
therefore, that the heights of curves depicting the acquisition of a
simple instrumental response vary directly with level of food
deprivation over a limited range.

PRIMARY SOURCES OF DRIVE

22 hours privation
(iV=12)

3 hours privation
(iV-12)

3 4 5 6 7 8
Successive four-trial blocks

10 11 12

Fig. 3:7. Mean running speeds of rats in a short straight alley as joint func-
tions of number of rewarded learning trials and of length of food deprivation.
{Adapted from Fredenburg, J 956.)

In attempting to interpret the results of studies such as these
one must proceed with caution. Performance may be better when
a response is learned following long deprivation, but this may
or may not mean that the manipulated variable (deprivation time)
is functioning "motivationally." Actually, stronger associations
or habits might be formed when an animal is hungry than when
it is not, and if behavior is assumed to be a joint function of
habits and drives, the observed disparities in performance might
be due to habit-strength differences, not to drive differences. Alter-
natively, such results could be due to inequalities in both drives
and habits or in drive level alone.

While few experimenters have studied the effects of a motiva-
tional variable upon the acquisition of instrumental responses, a
great many studies have been performed in which such variables
were manipulated during the extinction of a response. Tlie prob-
lem under investigation in these experiments is that of how differ-
ent drive levels affect resistance to extinction resulting from non-
reward.

Experimental designs evolved to shed light on this problem have
assumed a variety of forms. In a number of these, all subjects are
given their initial training under the same motivational conditions.

80 THE MOTIVATION OF BEHAVIOR

Then, when their performance has reached or approached a stable
level, they are divided into two or more subgroups that are given
extinction trials under different motivational levels. In general,
studies employing this design (Heathers and Arakelian, 1941;
Perin, 1942; Saltzman and Koch, 1948; Horenstein, 1951) have
found that resistance to the extinguishing effects of nonreward
varies directly with deprivation time. With this design, however,
it is difficult to draw clear-cut conclusions about the relative roles
of habits and drive. This is because some of the groups are extin-
guished under the same deprivation conditions as those obtaining
during learning, whereas others are extinguished under different
deprivation regimens. Thus for some groups the intensity or quality
of the internal stimuli (S^s or Smvs) would be the same during
acquisition and extinction, whereas for others marked changes
might be expected. Altering the nature of these internal cues
would affect performance if the response under study had become
associated with those cues. If all animals were trained under a
high level of deprivation and separate subgroups were extinguished
under that and lower levels, an observed decrease in resistance to
extinction with declining motivation could be explained as an
instance of weakened internal cues. Such an interpretation would
qualify as associative rather than motivational.

The difficulties attending this design might be overcome by
giving all subjects extensive preliminary training under all the
levels of deprivation to be used during extinction. Provided that
habit strength does not depend on drive, such diversified training
should result in the instrumental response's becoming conditioned
equally to the several intensities or kinds of internal deprivation-
produced cues. Response strength manifested by subgroups extin-
guished at different privation levels would then not be distorted
by uncontrolled variations in associative strength. To the best of
our knowledge, no one has performed an experiment in precisely
this way. The nearest approximation is a study by Cotton (1953).
His subjects (rats) were first given preliminary training in a
straight runway under a 22-hour deprivation schedule. This was
followed by extensive training under deprivation times of 0, 6, 16,
and 22 hours, administered in a random order within each four-day
period. When running-speed asymptotes had been reached, further

PRIMARY SOURCES OF DRIVE 81

tests were carried out under the four different privation schedules.
The measure of response strength was running speed on rewarded
trials rather than resistance to extinction. When all test trials
were counted, running time was found to decrease as a linear
function of deprivation time. Quite a different function was ob-
tained, however, when the trials on which animals made compet-
ing responses, such as stopping to sniff, were excluded. Under
these conditions performance was but little affected by deprivation,
a result that underscores the importance of the particular measure
used in studies of this kind.

Some of the difficulties inherent in shifting from one motiva-
tional level during acquisition to another during extinction can
also be overcome by the kind of design used by Yamaguchi ( 1951 ) .
He employed five different deprivation times (3, 12, 24, 48, and 72
hours), one group of animals being both trained and extinguished
at each of these levels. Thus for each group, the motivation-
variable-produced stimuli (Sj/i-s) would be the same during both
learning and extinction. The response was that of pressing a lever
in a modified Skinner box, and the basic measure was the number
of responses made during a series of extinction trials to a criterion
of no responses in two minutes. As in previous studies, Yamaguchi
found that the mean number of responses in extinction increased
progressiveh" as the period of privation was lengthened. He ob-
tained means of 14.9, 15.8, 24.4, 32.2, and 40.0 responses for 3,
12, 24, 48, and 72 hours of hunger, respectively. Considerable
confidence can probably be placed in the stability of these results,
since the number of subjects in each of the five groups was quite
large, ranging from 36 to 66. The possibility still remains, however,
that associative strength depends, to a degree, upon level of motiva-
tion, and this would operate artificially to enhance the apparent
energizing effects of extended deprivation times. Moreover, differ-
ent results might have been obtained had the distribution of
extinction trials been the same as that of the acquisition trials.
In Yamaguchi's study extinction trials were massed, whereas the
learning trials were given in a quasi-distributed manner: four trials
per session, one session every five days.

As we have seen in our discussion of behavior strength as a
function of time of deprivation, difficult problems arise in design-

82 THE MOTIVATION OF BEHAVIOR

ing experiments on motivational variables, and since known inves-
tigations differ so widely with respect to subjects, responses, and
procedures, no single function can be held to be uniquely repre-
sentative of the relation between response strength and deprivation
time. Nevertheless, three conclusions for which there is reasonable
support may be tentatively stated. First, consummatory activities,
as well as the speed and resistance to extinction of responses
instrumental to consummation, tend to be enhanced by privation.
The effect of any specific deprivation period is also a function,
however, of such variables as the number of deprivation cycles
preceding the test, the animal's body weight, and the relative
familiarity of the testing environment. Second, curves depicting
the course of acquisition of instrumental responses as a function
of different deprivation levels tend to diverge during early learning
trials and to maintain a constant separation with extended practice.
Third, when performance is measured at its asymptote, behavior
strength is found to rise rapidly after satiation (or shortly there-
after) up to privation times of about two hours and then to
increase at a slower rate to periods of about 48 to 72 hours. With
intervals longer than this, behavior strength declines, presumably
as a consequence of inanition.

Instrumental Behavior and Level of Noxious Stimulation. While
food and water deprivation are the most commonly manipulated
motivational variables, our treatment of primary sources of drive
would be incomplete without some mention of the ways in which
intense stimuli may motivate instrumental responses.

In experimental work with animals, electric shock, because of
the relative ease with which it can be applied and controlled, has
been used more frequently than any other noxious stimulus. Ex-
periments have also been reported, however, in which loud sounds,
bright lights, intense olfactory and gustatory stimuli, and exces-
sively hot or cold environments have been employed as motivators.

As the student of learning is well aware, noxious stimuli are
used as unconditioned stimuli (UCSs) in studies of classical
(Pavlovian) aversive conditioning with both human and animal
subjects. Unfortunately, UCS intensity has seldom been system-
atically varied in animal experiments, and it is necessary, therefore,
to rely upon instrumental conditioning studies for information as

PRIMARY SOURCES OF DRIVE 83

to the motivational consequences of this variable. In Chapter 1,
the role of \]CS> intensity in classical aversive conditioning with
human subjects will be discussed.

In research on noxious stimuli as motivators of instrumental
actions, the situation is commonly arranged so that the subject,
by performing either a previously established response or by learn-
ing a new one, can escape from or terminate the noxious stimulus.
Proficiency in the performance of the designated response is then
measured at different levels of the noxious stimulus. Experiments
by Amsel (1950cz) and by Campbell and Kraeling (1953) serve
as examples of this general procedure.

In AmseFs study, two groups of hooded rats were given practice
in escaping from a section of a straight alley with an electrified
floor into a wooden-floored goal box. The two groups were satiated
with food and water at the time of these trials and, save for the
fact that one group was given weak shocks and the other strong,
were treated identically. The performance of the two groups was
evaluated by measuring the time they took to traverse the middle
4-foot section of the 60-inch alley. Figure 3:8 presents the median
running times for the strong-shock and weak-shock groups as a
function of 10 successive training trials administered during a
single day.

From the data in Fig. 3:8 it appears that although the animals
were not motivated by either hunger or thirst, the shock provided
a strong drive to activate running, and its cessation following
escape from the alley functioned as a reinforcing agent. More-
over, although both groups quickly acquired the new skill of
running to escape shock, the strong-shock group maintained a
consistently higher performance level (shorter running times)
than did the weak-shock group, Amsel tried to prevent the rats
from becoming negatively adapted to the shock by gradually
increasing its strength, but the tendency for both curves to rise
toward the end of a second day of training (not shown in Fig.
3:8) may indicate the occurrence of such adaptation.

The experiment of Campbell and Kraeling (1953) also shows
that the performance of rats in an escape situation is facilitated
by increasingly strong electric shock. Acquisition curves obtained
from three groups of animals that were escaping from different

THE MOTIVATION OF BEHAVIOR

12 3 4 5 6
Trials

Fig. 3:8. Median running times over trials for two groups of rats (N = 28
per group) given weak and strong shocks in a short straight-alley maze.
{Adapted from Amsel, 19S0a.)

levels of shock into a neutral goal box are reproduced in Fig. 3:9.
As these data indicate, the strong shock provided a more power-
ful drive for running, even on early trials, than did the weaker
shocks. In commenting on these and other groups run in their
investigation, Campbell and Kraeling state that on the initial trials
most of their subjects ran at a speed proportional to the intensity
of the shock. Nevertheless, the performance level of the three
groups differed but little by the time training was terminated.
On the basis of statistical analyses of these data, Campbell and
Kraeling concluded that rate of acquisition varied significantly
with amount of shock reduction but that final running speed did
not. They suggest that differences in response strength might
have been observed if a different measure had been used. If all
animals were running at or near their maximum speeds by the

PRIMARY SOURCES OF DRIVE

80 -

70 -

60 -

£ 50

40 -

30 -

400 volts\ ^o °^

^^^ ^ -^^-^——H

-"-^.--^ f

.^ i

^^"^300 /

A'^ volts

- n

/^200 volts

.y^

1 I 1 1

2-3

4-6 7-9

Trials

10-12

13-15

Fig. 3:9. These curves show the effects of different intensities of electric
shock on the speeds with which rats run to escape shock. Speeds for trial 1
are plotted separately to indicate the effect of shock on running speed prior
to the first successful escape experience. {From Campbell and Kraeling, J 953.)

end of practice, an artificial ceiling would be imposed on mag-
nitude of response.

To the writer's knowledge, the only investigation in which a
functional relation has been obtained between time of air depriva-
tion and performance is one by Broadhurst (1957). Since there
seems to be little question of the noxious nature of stimulation
resulting from interference with breathing, studies of its motiva-
tional properties quite properly deserve mention at this point. The
curve of Fig. 3:10 reproduced from Broadhurst's paper is, in large
measure, self-explanatory. It shows that the speed with which rats
swam under water through a short straight alley increased
progressively as the period of air deprivation prior to their release
was increased. Broadhurst used 20 rats in this portion of his ex-
periment, all of them being tested at each of the several air-
deprivation intervals. The subjects had had preliminary practice
in underwater swimming before these tests, and it seems likely.

THE MOTIVATION OF BEHAVIOR

therefore, that this curve provides a reasonably accurate picture
of the effects of air deprivation upon performance level at or near
the learning asymptote. The slight drop of the curve from the
20- to the 25-second point may be, as Broadhurst suggests, a con-
sequence of anoxia at the longest deprivation interval.

Incidentally, a stimulus such as electric shock may function as
a primary source of drive for the learning of new responses even
when applied directly to the brain rather than to peripheral
sensory surfaces. For example, in a study by Cohen, Brown, and
Brown (1957) stimulation of the hypothalamus in cats served to
motivate the learning of both instrumental escape and anticipatory
(conditioned) avoidance responses. Central stimulation thus ex-
hibits functional similarities to the traditional grid-shock UCS, a
finding confirmed by Bower and Miller (1958), and by Roberts
(1958), among others. The two methods of shock administration
differ, however, in that central shock seems to lead directly to a
motivating emotional state without arousing the pain that grid
shock produces. Intracranial shock may thus be an instance of a
motivational variable having minimal directive (stimulus) conse-
quences and maximal motivating effects. This has important im-
phcations for the interpretation of experiments in which the onset

0.42

0 5 10 15 20

Seconds of air deprivation prior to test

Fig. 3:10. The speed with which rats swim through an underwater maze
as a function of the length of an immediately preceding period during which
they have been deprived of air. The ordinal values are reciprocals of swim-
ming times in seconds. {Adapted from Broadhurst, 1957.)

PRIMARY SOURCES OF DRIVE 87

of central shock proves to be reinforcing rather than punishing,
but v^^e shall postpone our discussion of this topic until Chapter 9.

It seems justified, from these and other studies (e.g., Miller
and Dollard, 1941; Kaplan, 1952; Ketchell, 1955; Spence, 1956),
to conclude that under certain conditions intense stimuli can
serve as a primary source of drive to motivate the performance
of instrumental acts. Moreover, responses followed by the cessation
of such stimuli tend to be learned, and the stronger the stimulus,
within limits, the better the performance. Nevertheless, it is well
to remember that the motivating and rewarding function of
strong external stimuli is always complicated by their capacity to
elicit specific responses. These reactions may be the correct ones
in a given experimental situation, or they may be wrong and in
competition with the correct ones, and their relative strengths
may vary with stimulus intensity. As an example, if weak grid
shock elicits responses of jumping and running in rats, oriented
escape reactions will perhaps be readily learned. But with intense
shock, crouching may become prepotent over running, and the
process of learning to escape may be retarded even though level
of motivation may be higher than before. Clearly, any conclusions
as to the motivating function of strong stimuli should, in principle,
include some reference to the associative (response-eliciting) prop-
erties of those stimuli.

Up to this point, our discussion of the effects of primary sources
of drive has dealt only with relatively simple situations and simple
responses. To round out the picture we turn now to the exam-
ination of a few selected investigations in which primary sources
of drive have been varied in choice and discrimination situations.

Behavior in Spatial Choice Situations as a Function of Primary
Motivational Variables. The simple T maze is one of the most
frequently used experimental situations in which animals must
learn to choose between two alternative goals or response se-
quences. Since the response to be learned is simply to turn toward
and approach a goal on either the left or right, the T maze is
sometimes described as a two-choice spatial situation. It is usually
distinguished, therefore, from a discrimination situation in which
the animal must approach a specific stimulus, such as a white
card, regardless of its spatial position.

THE MOTIVATION OF BEHAVIOR

Choice of the correct side of the T maze may be induced in a
number of ways. The animal may be rewarded on one side but
not on the other; he may be given more frequent rewarded trials
or larger rewards on one side; or he may be rewarded on one side
and punished on the other. Sometimes animals are permitted to
correct an error that has led to punishment or nonreward (cor-
rection method), and sometimes they are simply removed from
the maze (noncorrection method). In all cases efficiency of per-
formance is evaluated by counting the number of choices of the
correct side during the course of learning, these data being sup-
plemented occasionally by measurements of running speed in the
stem or to the correct and incorrect sides.

Another practical device for studying the learning of choice reac-
tions is the apparatus described by Logan (1952). The schematic
drawing in Fig. 3:11 shows that the principal features of the
apparatus are a starting box and a choice compartment. When the
guillotine doors are opened, the rat must leave the starting box,
enter the choice compartment, and touch one of the two small
bars projecting out into the compartment from the back wall.
When the correct bar is touched, a small pellet of food drops

Opaque barrier

Scale of inches

Glass door

Metal door

Starting box

Fig. 3:11. Floor plan of apparatus
used to investigate choice behavior
in rats. At the experimenter's dis-
cretion, either or both sides of the
choice compartment can be made
distinctive by svi'itching on lights
above each of the bars. Response
speed is obtained by measuring the
time that elapses between the open-
ing of the glass door and the mo-
ment at which the rat touches a
bar. {Adapted from Logan, 1952.)

PRIMARY SOURCES OF DRIVE 89

into the food cup below the bar and the bar is withdrawn from
sight. As in the T maze, performance may be evaluated either
from number of correct choices or from response speed. Speed
is determined by measuring the time between the opening of
the doors and the touching of the bar. While the T maze and the
two-bar choice box differ in many respects, the same principles
of learning and motivation may be applicable to both. Certainly
the problem of immediate concern with both is whether proficiency
in the execution of choice reactions is meaningfully and lawfully
related to variations in the strengths of primary sources of drive.

Systematic experimental studies of choice behavior as a function
of drive are relatively scarce, but the weight of evidence points
to the conclusion that selection of the correct side is unrelated to
time of food deprivation. This generalization seems to be restricted,
however, to experiments in which the animals are not permitted
to rectify their errors (noncorrection method) and in which they
are forced to go to the incorrect (nonrewarding) side as often as
they are permitted to go to the correct side.

The study we use to illustrate this conclusion is one whose
different aspects have been reported by Teel and Webb (1951)
and by Teel (1952). Four groups of albino rats, 21 per group,
were trained in a single-unit T maze under food-deprivation
periods of 1, 7, 15, and 22 hours, respectively. Of the four trials
given each subject on each of 14 days, two were free-choice trials
and two were forced. On free trials both choice-point doors were
open, whereas on forced trials only one was open and access to
only one arm of the T was permitted. This procedure was used
to equate the number of reinforced and nonreinforced runs for
all animals. The arbitrary criterion of learning was eight correct
choices in succession.

Analyses of the data revealed that the mean number of trials
required by the four groups to reach the criterion of learning did
not decrease with increasing time of food deprivation. In fact, a
small but consistent trend in the opposite direction was obtained.
It was also found, as reported in the Teel and Webb paper, that
when trials were administered under near-satiation conditions, level
of performance was not significantly reduced. Again, this indicates
that performance in a choice situation, as estimated by correct

90 THE MOTIVATION OF BEHAVIOR

reactions, is unrelated to deprivation time. Data tending to corrob-
orate those of Teel and Webb have been reported by Loess
(1952), who used the two-bar choice box; by Champion (1954),
who employed a T maze; and by Carlton (1955), who trained his
subjects on an elevated Y-like maze.

These studies are consistent in finding choice proficiency to be
unrelated to level of deprivation, but other experiments in which
different procedures were followed have produced positive results.
For example, Ramond (1954) found, with the two-bar choice
box, that rats under 22 hours of food deprivation performed
significantly better than others under 4 hours of deprivation. His
subjects, however, were given twice as many reinforced trials to one
bar as to the other, and his performance measure was the per-
centage of times the more frequently rewarded bar was chosen.
Thus, frequencies of left- and right-bar choices were unequal,
which was not the case in studies with negative findings.

Additional evidence for a positive relation between strength
of motivation and behavior in simple and multiple choice situa-
tions may be found in the investigations of Siegel (1943), Tolman
and Gleitman (1949), Miller (1948ib), and Davenport (1956). In
all of these studies, however, as in Ramond's, the procedures were
such that the number of responses made to the correct and incor-
rect sides of the apparatus were either purposely made unequal
or were permitted to become unequal during the course of learn-
ing. As Spence (1956) and Ramond (1954) have pointed out,
one would not expect a variation in drive to affect choice behavior
unless the training methods produce differences in the strengths
of the habits corresponding to the two reactions. If the habits are
kept equal by the use of forced trials, drive differences arising
from changes in deprivation should not affect percentage choice.
The explication of the details of this prediction is rather lengthy,
and hence it will be postponed to the next chapter, where we
shall be concerned more directly with such matters.

Behavior in Discrimination Situations as a Function of Primary
Motivational Variables. As we have noted, it is customary to apply
the term discrimination learning to situations in which an animal
must acquire a specific response to an environmental cue having
no fixed spatial location. For instance, in a black-white discrimina-

PRIMARY SOURCES OF DRIVE 91

tion the positive stimulus is usually presented equally often in the
right and left positions in the apparatus. Hence, success in the
performance of the discrimination cannot be achieved unless
tendencies to prefer one side or the other are overcome. It is in
this principal respect that discrimination situations differ from the
spatial-choice problems we have just examined.

It is somewhat regrettable, from the point of view of under-
standing phenomena in this area, that investigations of the role
of motivation in discrimination learning are even less plentiful
than those concerned with simple choice reactions. Perhaps the
earliest relevant study is one by Yerkes and Dodson (1908), who
examined the relation of shock intensity to performance on visual-
discrimination problems of varying levels of difficulty. One of the
principal findings of this investigation was that performance on
easy discriminations tended to improve progressively with shock
intensity. But on difficult problems performance was poorer with
weak and with strong shocks than with shocks of intermediate
strength. This latter finding led them to propose a general prin-
ciple, since known as the Yerkes-Dodson law, to the effect that
there is an optimal motivational level for learning, which tends
to decrease as problem difficulty increases. Subsequently, Dodson
(1917) performed a follow-up experiment, in which rats were
taught a light-dark discrimination under several intensities of
hunger. The results obtained were coordinate with those of the
first study, since performance improved for values of food depriva-
tion up to 41 hours but became worse when the period was ex-
tended to 48 hours. Shock was also used for some animals in the
second study and was found to be most efficacious when neither
too strong nor too weak.

More recent studies of the relation of hunger to performance
in a discrimination situation are those by Dinsmoor (1952), by
Eisman (1956), and by Eisman, Asimow, and Maltzman (1956).
In Dinsmoor's investigation, rats were trained initially to press a
bar while starved down to 85 per cent of their normal body weight.
Then they were given further training in making the discrimina-
tory response of pressing the bar in the presence of a light and not
pressing it in the dark. Final tests were carried out with satiated
animals and with animals maintained at 100, 95, 90, 85, 80, and 75

THE MOTIVATION OF BEHAVIOR

per cent of normal weight. The tests with percentage weights of
80 and 75 represent a greater food need than that in effect during
original training, and the weights of 90, 95, and 100, as well as the
satiated condition, constitute weaker need values. The results of
the study are presented in Fig. 3:12. Of principal interest here is
the finding that the difference between the number of responses
to the positive stimulus (light) and negative stimulus (no light)
increased as hunger was heightened above the level used in train-
ing. Thus an increase in hunger, even after fairly extensive training
has been carried out, seems to lead to improved proficiency in
responding differentially to the positive and negative cues. Dins-
moor's data also indicate that differences between responses to
the two cues decrease as hunger is reduced, with the total number
of responses declining also. Dinsmoor's animals were trained, as
has been pointed out, at only a single level of deprivation, a
procedure that may have permitted associative (habit) as well as
motivational (drive) variables to affect his results. As we shall see
in the next chapter, associative factors provide a reasonable basis
for explaining the convergence of the curves as drive level is re-
duced but not their divergence as drive is increased. Dinsmoor's

Positive-
stimulus

^ 1000

c
o

CL
I/)
<U

1 500

<,-——^

, Negative
/stimulus

1 1 1

Sat.

100 95 90 85 80 75
Per cent normal body weight

Fig. 3:12. Data obtained in a study of the effects of varying degrees of
hunger on behavior in a discrimination situation. The ordinate shows the
number of lever-pressing responses made to the positive stimulus (light)
and negative stimulus (no light) during a 90-minute test period. The initial
points of the curves were obtained following a period of continuous feeding
(satiation). {From Dinsmoor, 1952.)

PRIMARY SOURCES OF DRIVE 93

experiment, therefore, seems to support the earher conclusion of
Dodson that the performance of a (simple) discrimination task
is facilitated increasingly by heightened hunger.

The remaining two experiments of Eisman, Asimow, and Maltz-
man and of Eisman provide further evidence for a positive relation
between discrimination-task performance and level of motivation.
In the first of these experiments, rats run in a black-white, Y-type
discrimination apparatus under 46 hours of food deprivation made
fewer errors and required fewer trials and reinforcements to reach
a learning criterion than did rats trained under 22 hours or 4
hours of deprivation. The 4-hour and 22-hour groups did not, how-
ever, differ significantly with respect to any of the three measures.
This rather unexpected finding was interpreted by supposing that
the rat's stomach takes about four hours to empty following eating,
and that after this further deprivation has no effect upon drive.
On these assumptions, the drive levels of the 4- and 22-hour groups
should have been identical.

But the fact that the 46-hour group was superior to both of the
others could not be explained without additional assumptions.
As a consequence, Eisman (1956) suggested that drive level also
depends upon the general severity of the deprivation schedule
under which the animal has been maintained. This variable which
he termed "hours of deprivation during a unit of time" is believed
to operate jointly with the more conventional factor of "time
since eating." Thus one is led to the expectation that if two groups
have been fed, say, 22 hours previously, but if one of them has
been maintained on a 46-hour deprivation schedule, it will have a
higher drive level than one maintained on a 22-hour schedule.
Eisman's experiment, designed specifically to test this hypothesis,
yielded clear confirmatory results. More directly related to the
topic under discussion, however, was Eisman's finding that high-
drive groups, defined by reference to the severity of the mainte-
nance schedule, performed more efficiently on a black-white dis-
crimination problem than did low-drive subjects.

It might seem justifiable, in the light of the evidence provided
by these studies, to conclude that, in general, performance in the
learning of a discrimination reaction improves as a function of
heightened motivation. But exceptions to this conclusion are to

94 THE MOTIVATION OF BEHAVIOR

be found in the performance of the 4- and 22-hour groups of the
Eisman-Asimow-Maltzman study and in an unpubHshed study by
Myers (1952). This latter investigator found no relationship be-
tween drive differences due to 4 and 21 hours of food deprivation
and performance in a T-shaped brightness-discrimination appara-
tus. Meyer (1951) has likewise found that in experiments with
monkeys rather large differences in time of deprivation do not
affect performance in a discrimination-reversal situation. Until
these negative findings have been satisfactorily accounted for, the
wisest course of action is to withhold final judgment. Further
studies of the relations of changes in primary sources of drive to
discrimination learning and of the conditions determining specific
outcomes are urgently needed.

Summary

The first sections of this chapter were concerned primarily with
the elucidation of various terms frequently encountered in discus-
sions of motivation. Initially it was observed that a distinction is
commonly made by theorists between motivating or activating
variables and variables that function to guide or direct behavior.
The terms drive, conations, emotions, and motivation have been
used in conjunction with members of the first class, and cognitions,
associative bonds, and habit strengths with the second class. Cer-
tain variables, however, are diflBcult to classify because they appear
to function both as directive cues that elicit specific responses and
as relatively nonspecific motivators. To reduce confusion it was
suggested that such variables be conceptualized as leading to two
intermediary processes having the nonoverlapping functions of
behavior direction and of motivation, respectively.

As a further step toward terminological clarity, it was proposed
that the use of the plural word "drives" be dropped in favor of
the singular form. This suggestion was supported primarily by the
contention that the so-called drives can be differentiated with
respect to their individual antecedent variables, but not in terms
of their postulated effects upon behavior. The phrase "multiple
sources of drive" is proposed in place of the word drives.

In accordance with current practice, a distinction was also made

PRIMARY SOURCES OF DRIVE 95

between sources of drive that achieve their motivating effects
through the operation of inherited homeostatic mechanisms and
those whose effects are largely contingent on prior learning. Tlie
former are here termed "primary sources of drive" and the latter
"secondary sources of drive."

The variables of food and of water deprivation have been studied
more extensively than any others by students of primary sources
of drive. And since intakes of food and water have often been used
as indexes of the resulting drive, the problem of how such intakes
are regulated becomes of considerable significance. Contemporary
thinking in this area places primary emphasis upon two factors:
oropharyngeal stimulation produced by chewing and swallowing,
and stomach distention. Each of these processes exercise some con-
trol over ingestion when operating separately, but neither is
alone sufficient to account for all consummatory behavior. Thus
stimulation of receptors in the mouth and throat serves as a signal
tending to produce cessation of eating or drinking, but unless
aided by additional signals from the stomach, the control of in-
gestion is incomplete. Circulating nutrients in the blood and
stored nutrients in the tissues probably also play a role, but the
specific mechanisms involved have not been clearly detailed.

In the context of our discussion of factors controlling ingestion,
it was observed that a distinction can be profitably made between
need and drive. This is because drive can apparently be reduced
by variables that cannot possibly modify a physiological need and
because a need arising from deprivation may continue to increase
until death, whereas the resulting drive increases and decreases
with time. It was also noted that the maintenance of this distinc-
tion may clarify the interpretation of experiments directed against
the drive-reduction hypothesis of learning.

The major portion of the chapter dealt with studies of the
effects of primary sources of drive upon consummatory behavior,
upon instrumental responses, and upon performance in discrimi-
nation situations. Concerning the relation of deprivation to oral
intake, available data indicate that consummatory activity in-
creases as a monotonic function of hours of deprivation up to
periods of the order of two or three days. Beyond these limits
ingestive activity declines, presumably as a consequence of general

96 THE MOTIVATION OF BEHAVIOR

bodily weakness. Estimates of drive strength based upon consum-
matory responses are complicated, however, by learned tendencies
to eat or drink at certain times and by the kind of deprivation
schedule in effect prior to the test.

Studies of the effects of deprivation upon responses instrumental
to securing food or water support the conclusion that the speed of
these responses and their resistance to extinction tend to be en-
hanced by privation. Learning curves for different privation levels
tend to diverge at a relatively early stage of practice and to main-
tain a constant separation thereafter, asymptotic performance
levels being directly related to deprivation duration from about
2 to perhaps 72 hours, but not beyond.

Noxious stimuli appear to provide a convenient primary source
of drive, and performance under a variety of conditions, especially
those in which escape responses are investigated, is directly related
to stimulus intensity. The fact that exceptions are common, how-
ever, indicates that noxious stimuli may sometimes elicit compet-
ing responses and may thereby lead to performance decrements.

Experiments on the learning of spatial-choice reactions, as in T
mazes, are relatively consistent in showing that performance is
unrelated to privation when the associative strengths of the alter-
nate response tendencies are deliberately held equal. When one
response is more frequently reinforced, however, choice behavior
becomes more proficient as motivation is increased. In visual-dis-
crimination problems the effects of drive variations remain ob-
scure, since high drive sometimes facilitates performance and
sometimes does not.

CHAPTER

iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii^

Motivational and
Associative Interpretations
of "Motivated" Behavior

From time to time in the preceding chapters we have noted that
changes in behavior following the manipulation of motivational
variables can sometimes be explained by other than motivational
concepts. Moreover, we have stressed the view that any experi-
ment purporting to demonstrate motivational effects is not con-
vincing unless the influence of other factors has been eliminated
or controlled. Altering a so-called motivational variable may in-
deed modify drive strength; but it may also lead to changes in the
intensity and kinds of effective stimuli, in degree of learning, or in
the physiological condition of the organism. Whenever variations
in factors such as these provide acceptable explanations for the
observed behavior the concept of drive may become expendable.
The question arises, therefore, whether a specific drive construct
is needed at all by an adequate theory of behavior. Certainly, if
drive adds nothing to the clarity of our explanations or to the
breadth of our understanding, the principle of parsimony would
argue for its complete elimination. This possibility is tempting to

98 THE MOTIVATION OF BEHAVIOR

motivational theorists, and in recent years it has claimed their
attention to an increasing degree. Our purpose in the present
chapter is to evaluate the relative merits of the views that dispense
with, and appeal to, a drivelike concept.

To facilitate the discussion we have chosen to describe the al-
ternative views by the terms motivational and associative. A mo-
tivational theory is one containing, in a role of central importance,
a unique construct to which a specific label, e.g., drive, may be
attached. It is a theory in which the behavior-determining prop-
erties assigned to the drive construct are clearly different from
those ascribed to cognitions, habits, excitatory tendencies, and
other constructs. Hull's theory, which we have already considered
briefly, is a good example of a motivational theory because it con-
tains the construct D and because of the special properties as-
signed to D. However, a motivational theory, by our definition, is
not one that appeals only to motivational mechanisms in attempt-
ing to explain behavior. Nor must it place primary emphasis upon
such mechanisms to the exclusion of others. It is simply one that
provides specific room for a motivating mechanism as well as for
associative, excitatory, inhibitory, and other mechanisms.

By way of contrast, the type of theory we shall label associa-
tive contains no reference to concepts having such motivational-
sounding names as drive, tension, or libidinal energy. An advocate
of this type of theory would maintain that a separate drivelike
intermediary is not needed, and that most, if not all, of the con-
sequences of motivational variables can be explained by invoking
altered associative connections and their complex interrelations. If
this kind of view is to prove successful in interpreting the known
facts of motivated behavior, while maintaining an adequate degree
of internal consistency, it must not introduce additional constructs
having drivelike titles. Moreover, the functional properties of the
constructs in an associative theory must be different from those of
the drive construct in a motivational theory. Renaming drive with
a nondrivelike title will not suffice. Actually, highly detailed and
specific versions of associative theories are just beginning to ap-
pear, but the basic position has gained increasing numbers of
adherents in recent years, and it deserves, therefore, serious con-
sideration.

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 99

Motivational Interpretations

Hull's Multiplicative-drive Theory. Probably the best known
and most influential motivational theory is that of Hull (1943).
As we have already seen in the previous chapter, Hull assumed
that behavior is determined primarily by two major intermediary
factors: the nonspecific drive state (D) and a variety of learned
and unlearned reactive or associative tendencies. Unlearned reac-
tive tendencies are denoted by the symbol sUr, and learned ten-
dencies (habit strengths) by sHr, the S and R subscripts being
used to indicate the functional association of stimuli and responses.
We have chosen to apply the label "multiplicative-drive theory"
to this conception because Hull assumed that drive combines
multiplicatively with any reactive tendency to yield a response-
determining resultant called excitatory potential (sEr). Expressed
in the form of a simple equation with the subscripts omitted,
E = D X H, where E stands for excitatory potential, D for drive,
and H for learned associative tendencies. If the reactive tendency
were unlearned, U would be substituted for H in the equation.
Within the system, excitatory potential, modified by inhibitory
and oscillatory processes, governs the speed with which a response
will be made, its resistance to extinction, its amplitude, and so on.

This assumption that behavior is determined by the combined
operation of drive and associative tendencies has several implica-
tions that should be made explicit. In the first place, since drive
is conceived simply as a numerical multiplicative factor, it clearly
cannot determine the direction behavior will take. Such direction-
controlling functions reside in the reactive tendencies which may
be loosely described as blueprints for action, or as the organism's
"know how." Moreover, D as a nonspecific activator has only the
one multiplicative function irrespective of the source from which
it springs.

A second implication of the multiplicative hvpothesis relates
to the distinction commonly made by learning theorists between
learning and performance, where learning denotes the unseen
hypothetical changes produced by training and performance refers
to overt behavior. This distinction has arisen from the observation
that organisms sometimes do not perform well or perhaps do not

100 THE MOTIVATION OF BEHAVIOR

perform at all even when there is no good reason to suppose that
they have forgotten the skills requisite to a successful reaction. In
Hull's system, learning is represented by habit strength, and per-
formance is determined by excitatory potential. On the assump-
tion that drive and learned reactive tendencies combine multi-
plicatively, it is possible for him to maintain that when either
D or H has a value of zero, the product of the two is zero (i.e.,
E = 0) and no overt behavior should be exhibited. Thus an or-
ganism can "know" what to do (i.e., H can be greater than zero)
yet never exhibit this "know how" in its performance if the value
of D is zero. Such a conclusion could not be reached if D and H
were assumed to combine by addition rather than by multiplica-
tion.

A further consequence of Hull's hypothesis is that one cannot,
from the manifest characteristics of overt behavior alone, make
accurate estimates of the relative contributions of the two inter-
mediaries, D and H, unless a number of tests are given under a
variety of conditions. Any particular response could be due to
the presence of a strong drive combined with a weak reactive
tendency or to a weak drive and a strong reactive tendency. When
behavior does become manifest, both determinants are assumed
to be present. But if no response is exhibited, one cannot decide,
without additional information, whether a reactive tendency is
present and drive is not, or the reverse. If a response suddenly
fails to be exhibited where formerly it was commonplace, Hull's
position would suggest that associative tendencies are present but
not drive, since the former are assumed to be relatively permanent
and not subject to abrupt fluctuations.

Practical Implications of the Multiplicative-drive Theory. From
Hull's assumptions as to the manner in which a variety of different
antecedent conditions (motivational variables) contribute to drive,
and from his hypothesis of drive as a multiplier, a number of prac-
tical implications may be derived. In so far as these implications
can be supported by research findings, the drive concept gains in
acceptability.

For one thing, the theory implies that any motivational variable
can, within unspecified limits, be substituted for another since the
D provided by each is identical. Thus if a response has a high prob-

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 101

ability of being elicited by a stimulus when a subject is hungry,
it should also be elicitable, though not necessarily at the same
strength, when a subject is thirsty or in pain. In principle, the
response could be either learned or unlearned, because both kinds
of responses are affected by D in the same way. If positive results
are obtained, the theory tends to be supported. But negative
results, as we shall see at a later point in this chapter, may be
inconclusive since the shift from one drive-producing antecedent
condition to another may have modified the cues to which the
response was attached,

A second implication of the theory is that variations in strength
of drive produced by changing, say, time of food deprivation,
should ordinarily alter the quantitative but not the qualitative
features of a response. Increasing D should increase the speed,
amplitude, frequency, and resistance to extinction of a response;
but its specific direction should remain unaltered. Here again,
either learned or unlearned behavior could be studied, but positive
results would be expected only if the motivational variable does
not lead to marked changes in stimuli (Smvs) in addition to
changes in D.

The drive theory also implies that a wide variety of responses
should be enhanced by an increase in drive and weakened by a
decrease in drive. This is an obvious expectation from the hy-
pothesis that D multiplies all reactive tendencies. This prediction
might be tested experimentally by determining which response is
most likely to occur in each of a variety of different situations
when the subject is operating under a given level of drive. An
increase in drive produced by suitable manipulations of the ante-
cedent conditions should ordinarily lead to the enhancement of
all of these maximum-probability reactions. Conversely, all such
reactions should be weakened when drive is lowered.

The Multiplicative-drive Theory Applied to Various Kinds of
Behavioral Data. In order best to understand how the multiplica-
tive theory functions in the interpretation of behavior, let us apply
it to some experimental data of the kind described in the previous
chapter. Later on we shall show, for purposes of comparison, how
an associative theory containing no concept of D might account
for the same results.

102 THE MOTIVATION OF BEHAVIOR

Consider first an experimental result such as that reported by
Koch and Daniel (1945). These investigators trained rats to
depress a bar to obtain food in a modified Skinner-box apparatus.
At the time the response was acquired, the subjects were motivated
by 22 hours of food deprivation, and upon the completion of
training, the response was strongly established. After being thor-
oughly satiated with food, however, the rats would scarcely press
the bar at all. The median number of responses elicited from the
32 subjects prior to a five-minute no-response interval was only
one. From this result, Koch and Daniel concluded that the effec-
tive response strength under conditions of satiation was essentially
zero.

Now according to Hull's theory, the strength of the reactive
tendency (habit) to press the bar is not weakened by the operation
of satiation. The animal has forgotten neither how to make the
response nor where the food is located. Rather, the failure to act
is a consequence of the marked decrease in D accompanying
satiation. Suppose the associative tendency is given an arbitrary
numerical value of 5, and the D resulting from 22 hours of
deprivation a value of 10. Multiplying these two quantities together
vields an excitatory tendency (E) with a value of 50. Since the
animal does react under these conditions it must be presumed that
E is above threshold. But if the drive is reduced through relatively
complete satiation to a value of, say, 1 unit, the product of D and
H is only 5, and the threshold may or may not be exceeded. If
the threshold lies above 5, the animal will not respond at all.
The less complete the process of satiation, the smaller the reduc-
tion in the value of D and the less the performance decrement.
In this way, then, the multiplicative-drive theory attempts to
account for the fact that an animal will display a previously
learned response when it has been deprived of food, and not,
or to a lesser degree, when satiated.

In addition to affecting D, the operation of satiation
doubtless eliminates some of the internal stimuli (Sj/rs) pro-
duced by food deprivation and may also introduce new cues
arising from stomach distention. As we have already observed in
our discussion of Horenstein's and Siegel's experiments (Chapter
3), these interoceptive cues may function importantly as determin-

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 103

ants of behavior under changing conditions of deprivation pro-
vided they have become associated with critical responses. A "pure"
motivational interpretation cannot appeal to such internal cues.
To do so would be to invoke a nonmotivational (associative)
explanatory mechanism. In the foregoing interpretation of the
Koch and Daniel results, no such associative mechanism was
found necessary.

As Koch and Daniel have pointed out, however, the almost
complete lack of responses by their animals is apparently incon-
sistent with Hull's assumption that other primary sources of drive,
e.g., sexual deprivation, also contribute to nonspecific D. Their
subjects had been deprived of all recent sexual experiences, yet
D, which should have resulted from this deprivation, seemingly
did not multiply the bar-pressing tendency when hunger was
eliminated by satiation. The theory asserts that D, considered
solely as a multiplier, is unrelated to the source from which it
springs. Consequently, a shift from one primary source of drive
to another should result in almost perfect response transfer, pro-
vided that the second primary variable contributes as much to D
as the first, and provided the change involves the introduction
of no competing tendencies. To explain instances in which a shift
from one deprivation condition to another results in little transfer,
other principles must be introduced. The animals of Koch and
Daniel were not, of course, shifted from one primary source of
drive to a second. Rather their D was reduced from a level pro-
duced by hunger plus sex (and other factors) to a level dependent
only upon sex and other factors. Studies of the effects of gona-
dectomy upon the activity level of rats (e.g., Richter, 1933) sug-
gest that sex does make some contribution to general drive level.
It may be, therefore, that in the Koch and Daniel study the
motivating effects of sexual deprivation were overridden by com-
peting tendencies to rest induced by the extreme degree of satia-
tion.

The effects of changes in deprivation conditions upon the
behavior of animals in a spatial-choice situation constitute a second
typical observation to which the multiplicative-drive theory may
be applied. As was noted in Chapter 3, the experimental proce-
dures used to study behavior in these situations may be such

104 THE MOTIVATION OF BEHAVIOR

that the number of responses to each of the alternatives is either
equal or unequal. The multiplicative interpretation yields a dif-
ferent prediction for each of these situations, and it is instructive
to consider them in turn.

Ramond's (1954) study with a two-bar choice box (see Chapter
3) provides us with an example in which twice as many rein-
forced trials were given to one bar as to the other, separate groups
of animals being trained under 22 and 4 hours of food deprivation,
respectively. If we assume, following Hull, that the strength of a
habit depends upon the number of times the response has been
reinforced, then the habits to approach and touch the two bars
in this experiment should have been unequal, and since both
habits are multiplied by the same value of D, the corresponding
excitatory tendencies should also have been unequal. Therefore,
the animal should choose the more frequently rewarded alterna-
tive, since the dominance of one mode of reacting over another is
assumed to be determined by absolute differences in the strengths
of the corresponding E values. Thus an observed dominance of
one mode of action over another is attributed by this theory to
differences in habits or reactive tendencies. Drive makes it possible
for the "know how" to be exhibited in overt performance, but it
does nothing to alter the ranks of two (or more) reactive ten-
dencies. Drive has no inherent capacity to steer behavior into one
channel rather than another.

To continue the analysis, let us consider the effect on perform-
ance in the two-choice lever box of differences in D. Here Hull's
assumption of a multiplicative relation between drive and habit
leads to conclusions quite in accord with Ramond's findings.
That is, the relative superiority of a stronger over a weaker excita-
tory potential increases as drive is strengthened, and is reduced as
drive is weakened. To see precisely how this works let us assign
an arbitrary value of 10 to the habit tendency to press the left
bar, and a value of 5 to the right-bar tendency that has been
reinforced less often. If the drive due to 23 hours of deprivation
is assumed to be 4 units, the resulting E values obtained by
multiplying Hs of 10 and of 5 by 4 are 40 and 20, respectively.
The absolute difference between these E values is 20, with the
left-bar excitatory potential the stronger. But suppose that when

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 105

the animal has been partially satiated the strength of D drops to
2 units. In this case, the absolute difference between the E values
is decreased to 10, since the E value for the left bar drops to 20
and that for the right bar to 10. If the number of correct responses,
i.e., depressions of the more frequently reinforced bar, is deter-
mined by the absolute difference between the two Es, performance
should become poorer under the partial satiation conditions. Con-
versely, of course, performance should improve as strength of
drive is increased. Because changes in drive intensity may thus lead
to either better or poorer performance, drive, in an indirect sense,
does have an effect upon the direction behavior will take. It exerts
this effect, however, not by virtue of a steering function inherent
in drive itself, but through its power to change the absolute differ-
ence between the strengths of competing excitatory potentials.

An experiment by Teel (1952), described in Chapter 3, illus-
trates the application of the multiplicative theory to behavior in
a spatial choice situation (T maze) where practice in performing
the two responses has been equated. In this study, it may be
recalled, four groups of rats were trained under food deprivation
times of 1, 7, 15, and 22 hours, respectively. Each rat was forced
to run to one side as often as to the other to equate practice in
running to the rewarding and nonrewarding sides. When perform-
ance was evaluated in terms of the number of trials required to
reach a learning criterion of eight successive correct choices on the
free trials, all four groups learned with equal facility.

This finding can be handled by the multiplicative theory, as
Spence (1956) has shown, provided one makes the assumption
that the forcing procedure which produced equal runs to the two
sides of the maze resulted in equal habit strengths to approach the
rewarding and nonrewarding arms. This assumption, it should be
noted, is not one that Hull would have made, since he believed
that habit strength is built up only on reinforced trials. Rather
it is the kind of assumption made by those who hold that mere
contiguity of stimulus and response is sufficient to produce learn-
ing (e.g., Guthrie, 1935). On the supposition that the habit
strengths are equal, the multiplicative theory would predict no
change in the relative frequency of correct and incorrect responses
as a function of differences in D, since the two E values obtained

106 THE MOTIVATION OF BEHAVIOR

by combining D with the two Hs would be identical regardless
of strength of D. This was the result, as we have seen, of Teel's
investigation. To explain the fact that animals do learn to perform
correctly under these experimental conditions, resort must be
made to other factors, e.g., anticipatory goal reactions, which are
unequal for the two sides of the maze (cf. Spence, 1956).

An additional and quite important implication of the multipli-
cative theory is that performance will be degraded by an increase
in drive whenever the reactive tendencies corresponding to incor-
rect responses are stronger than those corresponding to correct
reactions. In Ramond's experiment, for instance, if the more
frequently touched bar were actually the wrong one from the
experimenter's point of view, intensification of D would lead to an
increase in errors and hence to a decline in proficiency. Moreover,
lowering D should tend to make performance better, a prediction
that is apparently unique to the multiplicative-drive conception.

In this context it should be noted that the theory cannot, in its
"pure" form, explain a drop in performance following a supposed
increase in drive if there is reason to regard the correct habit as
stronger than the wrong one. When such results are obtained,
appeal is customarily made to the presence of competing ten-
dencies, a solution that is essentially associative in nature. This is
legitimate, of course, provided independent evidence can be
presented to indicate that the motivational variable which led
to an increase in D also led to an increase in the strengths of
interfering habits or to an increase in the intensity of stimuli
associated with such habits.

In summary, a motivational theory such as Hull's explains
certain relatively simple behavioral phenomena by introducing
an explicit construct of drive (D). This factor is assumed to com-
bine in a multiplicative fashion with habitual or instinctive reac-
tive tendencies (Hs or Us) to yield the excitatory potentials (Es)
of which behavior is said to be a more or less direct function.
When a specific overt reaction occurs as a consequence of an
operation such as food deprivation and does not occur following
satiation, the multiplicative theory holds that D has been raised
by deprivation and the value of the corresponding E has been
elevated above a threshold. Since D itself is not regarded as a

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 107

director of behavior but only as a broadly acting energizing process,
the occurrence of one response rather than another is explained
as being due primarily to differences in the strengths of simul-
taneously present reactive tendencies. When two or more such
tendencies are present, an improvement in performance following
an increase in deprivation is accounted for by the assumption that,
as D is increased, the absolute difference between the strengths
of the stronger correct E and the weaker competing E is increased.
But if the wrong habit is dominant prior to the increase in drive,
then raising the level of D should lead to poorer performance.
When the correct tendency is believed to be dominant initially,
but increased D is followed by a decline in performance, it must
be assumed that other, incorrect habits have displaced the correct
ones at the top of the hierarchy of habits.

Other Motivational Theories

Hull's theory has served as the focal point for our discussion
of motivational interpretations because it is perhaps more widely
known than others and has been carried to a relatively high level
of systematic refinement. Numerous other theories, however,
might also qualify as motivational interpretations, and although
we cannot take time to show how each might be applied to experi-
mental data, a brief mention of several is appropriate at this point.

In its principal elements, Spence's (1948, 1956) theory parallels
Hull's since the construct of drive (D) is retained along with
the assumption that it multiplies associative tendencies. Spence
holds that drive is a direct consequence of deprivation, but that
variables such as unconditioned stimuli (for instance, air puffs
in eyelid-conditioning experiments) and electric shocks affect
drive indirectly via the elicitaiion of an emotional response. Since
drive strength is taken as proportional to the vigor of this response,
such descriptively different factors as individual differences in
emotional reactivity to stress, the number of prior exposures to
aversive stimuli, and stimulus intensity can all affect drive and
hence performance (Spence, 1958).

Spence's interpretation also departs somewhat from Hull's in
placing more emphasis upon a second motivational construct

108 THE MOTIVATION OF BEHAVIOR

termed incentive motivation (K). As a behavior determinant, K
is assumed to combine with D in an additive manner, and Hke D,
therefore, it multiphes reactive tendencies. The variables determin-
ing K, however, are different from those upon which D depends.
Specifically, K is a function of magnitude of reward, of the number
of times the reward has been experienced, its palatability, the
length of the chain of responses, and perhaps also of the time
between the initiation of the response and the receipt of the
incentive. As to its underlying mechanism, K is the consequent
of the occurrence of a classically conditioned partial consum-
matory goal response (r^). After an animal has been repeatedly
fed in a given situation, it will tend to make anticipatory chewing
and salivating responses prior to obtaining food. It is these
responses (r^s) in anticipation of reward that provide the mo-
tivational increment described as K. Because these responses are
largely learned, they qualify as secondary sources of drive and
will be discussed in greater detail in Chapter 5.

Other writers who have followed Hull in their treatments of
motivation (e.g.. Brown, 1953cz; Farber, 1954ci; Taylor, 1956^:;
Amsel, 1958) are readily identified as motivational theorists, but it
is more difficult to categorize those who fall outside the Hullian
tradition. Within Tolman's (1932) early system, for example,
demands (sometimes identified with purposes) tend to be equated
to drives or motivations and to occupy the status of intervening
variables. But demands, or at least drives, though they are initiated
by physiological conditions, consist in part of a sign-gestalt readi-
ness (a sort of perceptual-idea tional process) which at other times
is said by Tolman to be a constituent of cognitions. Thus Tolman
is a motivational theorist in postulating intervening variables
named drives or demands^ but the behavior-determining properties
of these variables are neither clearly specified nor carefully differ-
entiated from those of cognitions. In certain of his latest formula-
tions, Tolman (1951, 1952) seems to have moved even closer to
a concept of a general, nondirective drive, since he introduces a
libido need that has no specific goals of its own but is capable of
controlling the energy that is available for the satisfaction of
other specific needs.

Duffy (1934, 1951, 1957) has insisted for many years that a

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 109

clear distinction should be made between the factors that deter-
mine the direction^ behavior will take and those that govern its
intensity. Her use of such terms as "arousal," "activation/' and
"energy mobilization," in reference to the dimension of behavior
intensity is clearly congruent with a concept of a nonspecific
motivating agency such as Hull's D. Moreover, she has also main-
tained that the concept of activation aids in erasing an unwar-
ranted distinction between "drives" and "emotions," a view that
can readily be related to theories of fear as a learned source of
drive (Miller, 1951) and to the view that diverse environmental
variables arouse drive through a mediating emotional response
(Spence, 1958). Malmo (1957, 1958) has taken a position much
like Duffy's, though he has tended, where she has not, to identify
arousal level with specific physiological variables.

Among physiologically oriented psychologists, Morgan (1943,
1957) might be described as having espoused a motivational inter-
pretation because of his concept of the central motive state
(c.m.s.), which he defines as the neural integrative activity under-
lying motivated behavior. On the antecedent side, the c.m.s. is
aroused by receptor-mediated stimuli and by chemical hormonal
conditions. The latter are believed to affect the c.m.s. by direct
stimulation of certain brain centers or in such indirect ways as by
the excitation of internal receptors. Regarding its functional prop-
erties, the long-perseverating c.m.s. appears to serve in part as a
nonspecific energizer, since it leads to general activity. It also is
capable, however, of "emitting" particular responses even in the
absence of specific stimuli, and can sensitize the organism to
respond selectively to certain stimuli rather than to others. Morgan
has listed four ways in which the c.m.s. might be reduced or
negated: (1) by removal of the initiating humoral or stimulus
events; (2) by the release of other "humoral messengers" having
the capacity to reduce the c.m.s. directly; (3) by the excitation
of sensory receptors; and (4) by the performance of certain acts
generated by the c.m.s. As we have already noted in Chapter 3,
Morgan is among those who hold that drive (c.m.s.?) may be
controlled by sensory stimulation. Items (2) and (3) and perhaps
(4) of the above list clearly reflect this view. Morgan has not,
however, presented a formalized treatment of the c.m.s. or applied

no THE MOTIVATION OF BEHAVIOR

it to experiments in which motivational variables have been ex-
pressly manipulated.

Stellar (1954) has accepted Morgan's c.m.s. as the central
nervous mechanism underlying drive and has suggested that it be
assigned a definite physiological locus in the hypothalamus. In
broad terms, motivated behavior thus becomes dependent upon
the level of activity in "certain excitatory centers of the hypo-
thalamus."

Lindsley (1951, 1957) and Hebb (1955), both of whom have
been struck by the nonspecific activating effects of excitation in
the ascending reticular system of the brain, might also be tenta-
tively listed among the motivational theorists. Hebb, in particular,
has identified the arousal function of the reticular formation with
a general drive state and has urged that the energizing function
be distinguished from the "S-R or cognitive functions that are
energized." Neither of these writers, however, has presented us
with a formal system in which the precise role of the arousal
mechanism has been delineated. A brief resume of physiological
findings bearing on the concept of a nonspecific drive, as cited by
these and other writers, will be found below in Chapter 9.

Finally it should be noted that concepts such as Freud's (1905)
libido, McDougall's (1917) instincts, and Woodworth's (1918)
drive suggest that these authors might be classified as early advo-
cates of motivational interpretations. By certain standards, of
course, these formulations may seem relatively unsystematic but
clear anticipations of many of our present-day views can be dis-
covered therein.

Associative Interpretations of the Role
of Motivational Variables

Although motivational theories clearly have merit, there are
other reasonable interpretations of the same phenomena which,
because they make no reference to a construct such as D, may be
described as nonmotivational or, more specifically, as associative.
The ideas basic to any associative view are well known to students
of motivation, and suggestive allusions to them may be found in
various sources. Fundamentally, an associative view would hold

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 1 1 1

that the manipulation of a motivational variable leads to changes
in the stimulus conditions effective at the time performance is
measured. The altered stimuli, in turn, would affect behavior
through changes in the number, relative strengths, or kinds of
associations between those stimuli and overt reactions. To the
author's knowledge, however, the specific steps involved in the
development and application of associative interpretations have
rarely been presented previously. The following discussion repre-
sents the writer's preliminary attempt to formulate and evaluate
several versions of an associative theor)^

An examination of existing hypotheses as to factors responsible
for variations in associative strengths suggests that three relatively
distinct versions of an associative theory might be developed. The
basis of the first is that motivational variables alter behavior be-
cause they lead to the addition or removal of critical stimulus
elements and hence to the modification of existing habit strengths.
In the second, motivational variables are assumed to operate by
changing the quality or intensity of critical stimuli. This, in turn,
produces variations in associative strengths in accordance with
the specific hypothesis of habit-strength generalization. In the
third version, central prominence is given to the likelihood that
new habits, capable of either facilitating or interfering with the
old are brought into effective action by the manipulation of motiva-
tional variables.

Attributing the Effects of Motivational Variables to the Addi-
tion or Subtraction of Critical Stimulus Elements. To illustrate
this first form of an associative interpretation, let us refer again to
the Koch and Daniel (1945) experiment wherein animals were
trained to press a bar under long-term deprivation and were then
tested in the same situation under short-term deprivation (satia-
tion). In this case, the empirical phenomenon to be explained is
the decrement in performance produced by satiation.

According to the view under consideration, the original training
may be thought of as producing associations between the bar-
pressing responses and a constellation of both internal and external
stimuli. This assumption is represented in the upper part of Fig.
4:1, where stimuli Si, S2, S3, and S4 represent internal cues, Se
represents all the external cues of the testing situation, and H

112

THE MOTIVATION OF BEHAVIOR

indicates the associative connection of all cues to the bar-pressing
response. When tests are conducted under conditions of complete
or partial satiation, some, though probably not all, of the internal
stimuli originally conditioned to the reaction would presumably
be eliminated. This possibility is expressed in the lower half of
Fig. 4:1, where only Si and S2 of the original set are indicated as
present. Now if we make the explicit assumption that the strength
of an existing habit is directly related to the number of elements
that are present from the originally conditioned stimulus aggregate,
then the deletion of S3 and Si should weaken the bar-pressing
habit, and level of performance should decline. In effect then,
this view holds that behavior occurs when stimuli appropriate for
its elicitation are present and does not occur when they are
absent. Thus, the facilitative effect of long-term deprivation as
contrasted with the effect of satiation is explained without refer-
ence to the energizing or multiplying effects of a general D.

This version of an associative theory cannot, however, easily
explain the facilitation of performance that might occur when
training has been carried out under short-term deprivation and
tests are then made under long-term deprivation. Loess (1952),
for example, found that speed of response in a two-bar choice
box increased when time of deprivation was shifted from a training

Internal cues due
to long-term -
deprivation

External cues —

Training conditions

Sj, S2, S3, S^

-H-

-R

bar pressing

Internal cues due
to short-term
deprivation

External cues

Testing conditions

bar pressing

Fig. 4:1. Elements of an associative interpretation of a decline in per-
formance following a shift from a long-term to a short-term deprivation
schedule. The loss of stimuli S3 and Si occasioned by the shift in schedule
is assumed to weaken the original associative tendency. The stronger habit
is indicated by H, the weaker by h.

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS

113

value of 3 hours to a testing level of 32 hours. The hypothetical
factors involved in this series of events are shown in Fig. 4:2. The
upper portion of this figure represents the assumption that the
response of approaching and touching a bar has become strongly
attached (H) to the combination of Si, S2, and Se- If at this point
the deprivation period is lengthened, stimuli S3 and S4 should be
added to the complex (lower part of Fig. 4:2). But if these new
cues have never been previously associated with the response of
approaching and touching the bar, their addition to the stimulus
complex cannot explain the observed improvement in performance.
In fact, as deprivation increases, the internal stimulus complex
should become progressively more unlike that present during
original training; and if we accept current views as to the role of
stimulus similarity in determining associative strength, we must
conclude that the tendency to make the original reaction should
become weaker, not stronger, as the difference between the original
and subsequent stimulus complexes is increased. This possibility
is indicated in the bottom portion of Fig. 4:2 by the lower-case h.
It thus appears that, granting the above assumptions, we must
conclude that this particular associative interpretation cannot

Internal cues due
to short-term ■
deprivation

External cues

Training conditions

■*■ Sj, S2

+
— ^ S^

>-H — R

approach and touch bar

Internal cues due
to long-term -
deprivation

External cues ■

Testing conditions

•Sj, Sj, S3, S4
+

^Sr,

>-h—R

approach and touch bar

Fig. 4:2. The application of an associative hypothesis to behavior observed
when original training is carried out under a short-term deprivation schedule
and subsequent tests under long-term deprivation. From these assumptions
alone it would be predicted that, as deprivation is enhanced, the addition of
S3 and Si to the stimulus complex would lead to a weakening of the association
and hence to poorer rather than better performance.

114 THE MOTIVATION OF BEHAVIOR

readily deal with the case where original training has been admin-
istered under short-term deprivation and improved performance is
then observed under long-term deprivation.

Special-purpose modifications of the theory might be introduced,
of course, so as to make it work more convincingly. Thus, one
might maintain that Si and S2 become more intense with extended
deprivation and are therefore more likely to exceed a threshold,
or that S3 and Si have the power, as the result of previous learning,
to facilitate rather than interfere with the approaching response.
Modifications such as these, however, raise additional problems,
which when carefully analyzed may convert the theory into one or
the other of the two forms to be described below.

Our first associative hypothesis can, however, deal adequately
with decrements in performance attending a shift from one need
condition to another. Here it would be presumed that a change
from thirst to hunger, for example, results in the partial elimina-
tion of the thirst stimuli associated with the response during
training. The fact that the response occurs at all under hunger
would be explained on the grounds that some of the original in-
ternal (thirst) stimuli were still present at the time of the tests
under hunger. The theory is also satisfactory as an interpretation
of the effects of deprivation upon the acquisition of simple instru-
mental responses (cf. Fredenburg's data. Figs. 3:6 and 3:7). One
would probably assume, in this case, that many more internal cues
are present under 22 than under 4 hours of deprivation and that
running can become more readily conditioned to a CS complex
consisting of many elements than to one with few components.
Thus performance differences would be explained in terms of
differences in habit strengths.

In attempting to extend this type of associative theory to choice
situations in which one or more associative tendencies are simul-
taneously present and in which the deprivation conditions are
altered, reference would again be made to the addition or deletion
of stimulus elements. If a rat has learned to turn right in a T maze
when strong need-produced stimuli are present, its performance
would be expected to suffer if the need were reduced through
satiation. On the other hand, if the reaction has been conditioned
to weak need stimuli, there is no clear-cut prediction from the

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 115

theory that performance will improve when the need is increased.
As in the case of the single reaction considered above, it is difEcult
to see how performance could be improved by suddenly adding
irrelevant internal cues to the conditioned-stimulus aggregate.

Motivational-variable Effects Ascribed to Changes in the
Strength of Habits in Accordance with the Principle of Stimulus
Generalization. In order to explain this second type of associative
interpretation, in which changes in motivational variables are
assumed to affect the intensity or quality of stimuli rather than
their mere numerosity, we must digress briefly to consider both
the empirical phenomenon of stimulus generalization and the
hypothesis of generalized habit strength.

The empirical phenomenon of stimulus generalization is nicely
illustrated in an experiment by Grice and Saltz (1950). In this
study rats were trained to traverse a short, straight black alley
to obtain food from a small compartment in the center of a
white disk located on the end wall of the alley. Seventy-five animals
were trained to approach when the disk had an area of 79 square
centimeters. Five subgroups of 15 animals each were then tested
on disk sizes of 79, 63, 50, 32, and 20 square centimeters, respec-
tively. These tests consisted of 25 opportunities to approach a
disk, but no food was given on any trial. The strength of the
approach tendency was measured by counting the number of
times, during 25 trials, that an animal approached and touched
the disk within 60 seconds. The results of these tests, reproduced
in Fig. 4:3 (dashed-line curve), indicate that the animals approach
disks that have never been used in training, but that as the size
of the disk decreases, the tendency to approach becomes progres-
sively weaker. The phrase empirical stimulus generalization denotes
the fact that responses do occur to stimuli even in the absence of
specific training with those stimuli. The phrase gradient of gen-
eralization refers to an observed decrease in response strength as
the test stimulus differs progressively from the one used in train-
ing.

An additional 60 animals were trained by Grice and Saltz to
approach and secure food from the stimulus with the smallest
area (20 square centimeters). Four equal subgroups were then
given extinction tests with disks having areas of 20, 32, 50, and 79

116

THE MOTIVATION OF BEHAVIOR

i.H

/
/
/

•

\-<^

•

1 . 1

< 1

32 50

Area of test stimulus in sq cm

Fig. 4:3. The results of an experimental study of the generalization of in-
strumental responses to stimuli varying in the dimension of size. The dashed-
Hne curve shows the number of extinction responses made by animals trained
with the largest stimulus and tested on the others. The solid-line curve de-
picts the performance of rats that were trained on the 20-cm stimulus and
tested on that and on larger ones. {From Grice and Saltz, 19 SO.)

square centimeters. Here also, stimuli never before associated with
approaching elicited the response, and, in general, the more deviant
the size of the test stimuli the less resistant the response to extinc-
tion. Data obtained from these animals are represented by the
solid-line curve of Fig. 4:3.

This, then, is the empirical phenomenon of stimulus generaliza-
tion: after a response has become associated with one stimulus,
it may be elicited, though often with a decrement, by other similar
stimuli in the absence of specific training with those stimuli.

In attempting to account for empirical data of this kind, Hull
(1943) made the assumption that whenever a response is condi-
tioned to a particular stimulus, associative strengths are automat-
ically built up to (generalized to) other similar stimuli. Moreover,
the greater the disparity between the conditioned and test stimuli
the weaker the generalized habit strength. The specific details of
his assumptions are shown in Fig. 4:4, where habit strength is
indicated on the ordinate and a stimulus dimension on the ab-

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS

117

scissa. The conditioned stimulus (CS) is shown to have the
greatest habit strength and the generahzed stimuh (Ss with nu-
merical subscripts) are shown to have progressively weaker habit
loadings as their distance from CS increases in either direction
along the base line. Hull assumed that the gradients of generalized
habit strength will always have the general form shown in Fig. 4:4
irrespective of the particular stimulus dimension, provided the test
stimuli have been located at subjectively equal distances along
the abscissa. It is important to note that these curves do not
represent experimental data but simply Hull's hypothesis as to the
way in which the strength of an association changes with variations
in the stimulus. It is an assumption advanced to explain the em-
pirical phenomena of generalization. This assumption and the
empirical phenomena must be kept separate in one's thinking if
confusion is to be avoided.

Although the hypothetical curves of generalized habit strength
differ considerably from the empirical curves obtained by Grice
and Saltz, the latter do show a fairly regular decline as the distance
between the conditioned and the test (generalized) stimuli is

'1 ^^ ^1

Stimulus dimension

Fig. 4:4. A graphic portrayal of Hull's (1943) hypothesis of generalized
habit strength. CS represents the stimulus to which a response has been con-
ditioned, and the habit strength at that point is assumed to be maximal.
The stimuli to the right and left of CS are assumed to be located at psycho-
logically equal distances along the dimension, and the corresponding values
of generahzed habit strength are shown as decreasing with increased distance
in both directions from CS.

118 THE MOTIVATION OF BEHAVIOR

increased. On Hull's hypothesis, the rats in the Grice-Saltz experi-
ment approach white circles of sizes other than that used in train-
ing because habit strength generalizes to those other stimuli. The
fact that the rats show less resistance to extinction as the test
stimuli depart increasingly from the one used in training is at-
tributed to the decline of habit strength with distance along the
stimulus dimension.

Bearing in mind the empirical phenomenon of stimulus gen-
eralization and the hypothesis of generalized habit strength, we
may now return to the associative interpretation to which we
alluded at the beginning of this section.

For expository purposes we shall consider an experiment by
Deese and Carpenter (1951) in which two groups of rats were
given 24 training trials in running down a short straight alley for
food. At the time of this training, one group had just been fed
wet mash for 1 hour, whereas the other group had not eaten for
22 hours. At the end of this training period the drives of the two
groups were reversed and 8 more trials were administered. Now
according to the associative view we are considering, the early
training should produce learned connections between the running
response and a stimulus complex consisting of environmental cues
plus the internal cues (Smfs). Turning first to the group trained
under 22 hours of deprivation, the theory predicts that when the
animals are shifted to near-satiation conditions their speed of
running should decline, since the internal deprivation cues, if still
present, should be much weaker and the generalized habit strength
should therefore be reduced. To make this clearer, let CS in Fig. 4:4
represent the internal stimuli arising in conjunction with 22 hours
of food deprivation, and let us assume that the running response
has been conditioned to these cues along with those of the alley.
The stimuli to the left of CS would then represent internal cues
corresponding to shorter deprivation periods. Since the strength of
the habit declines with distance on the stimulus dimension from
CS, the strength of the tendency to run should also weaken as
time since eating is reduced. The results of the Deese and Car-
penter study are presented in Fig. 4:5, where it may be seen that
the performance of the high-drive group (solid-line curve) did
decline when deprivation was made less severe, though the drop

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS

119

is not extreme and performance becomes quite variable. This find-
ing is thus consistent with the associative hypothesis under exam-
ination, but it also supports a drive theory such as Hull's.

An associative interpretation involving the assumption of gen-
eralized habit strength can thus explain a decrement in perform-
ance following a shift from long- to short-term deprivation. It is
an explanation that stresses the role of stimulus variation in the
determination of associative strength. In this instance, however,
the stimulus is presumed to be internal rather than external, as in
the Grice-Saltz experiment. Unlike the preceding associative inter-
pretation, this one does not appeal to changes in the number of
associated stimulus elements. Nor does it, like the multiplicative-
drive theory, require the postulation of a general drive factor. In
Chapter 3 an associative interpretation of precisely this kind was
discussed as a possible alternative to the multiplicative theory in
connection with Horenstein's experiment, though we did not, at
that point, introduce the concept of stimulus generalization. By
its very nature, incidentally, this associative view is applicable
only to studies in which deprivation is shifted from one level to

0.55

qO.50

0.45

0.40

0.35

• • High drive-low drive

o o Low drive-high drive

."^^ k

12 16 20 24 2 4 6 8
Trials (reinforcements)

Fig. 4:5. Starting-speed scores for two groups of rats trained under long and
short food-deprivation schedules, respectively, for 24 trials and then given
8 more trials with the deprivation durations reversed. {Adapted from Deese
and Carpenter, 1951.)

120 THE MOTIVATION OF BEHAVIOR

another or to those in which one need is substituted for another.

From the relations depicted in Fig. 4:4, it may be seen that our
second associative formulation, like the first, cannot easily explain
the marked improvement in performance shown by the Deese
and Carpenter animals that were trained initially under near-satia-
tion conditions and were then shifted to the 22-hour deprivation
regimen (dashed-hne curve in Fig. 4:5). If CS in Fig. 4:4 now
represents the internal cues characteristic of near-satiation at the
time of initial training, stimuli to the right of CS would correspond
to cues attending deprivation periods of greater and greater dura-
tion. As these stimuli become increasingly different from CS the
corresponding habit strengths decline, and performance should
become poorer. The unelaborated hypothesis of generalized habit
strength, therefore, leads to the prediction that performance will
decline regardless of whether the deprivation period is increased
or decreased from that used in original training. Deese and Car-
penter, however, found that performance improved immediately
when deprivation duration was increased. Other experimenters
have also observed that a sudden increase in deprivation leads to
improved performance (e.g., Skinner, 1938; Loess, 1952; Hillman,
Hunter, and Kimble, 1953; Davis, 1957; Barry, 1958; and Bass,
1958), though the magnitude of the effect varies with the response
being measured and with situational variables.

The associative hypothesis could be modified to take account
of the Deese and Carpenter findings by changing the basic assump-
tions concerning the form of the generalization gradients. Specifi-
cally, one might assume that when the stimulus dimension is one
of intensity rather than of quality, the gradient of generalized
habit strength actually rises progressively from the point of condi-
tioning {CS in Fig. 4:4) to other points at the right of the dimen-
sion. But this would lead to the conclusion that the generalized
habit strength for an intense stimulus is greater than the condi-
tioned habit strength for a weaker value of the same stimulus even
though the former has never been experimentally associated with
the response. To the writer's knowledge, no one has seriously
championed such an interpretation, but when test (generalized)
stimuli are more intense than the conditioned stimulus, the
empirical generalization gradients tend to be flatter than when

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 121

the test stimuli are weaker than the conditioned stimulus (cf.
Miller and Greene, 1954). A consistent associative theorist could
not, of course, assume that strong stimuli have inherently more
driving power than weak ones, since this would constitute a moti-
vational rather than an associative conception.

Experiments in which an associative tendency is established
under one motivational variable and performance is measured
under another can also be interpreted in terms of habit-strength
generalization. Here the underlying supposition is that the internal
stimuli attending the substituted variable are similar to those
present when the associative tendency was first established. Thus
it might be supposed, with some reason, that hunger, thirst, and
nausea all involve similar visceral stimuli. From the postulate of
generalized habit strength the conclusion is reached that if a
response has been learned when an animal is thirsty, the same
response will be evoked, but with a decrement, when the animal
is made either hungry or nauseated. Note once more that this
associative explanation of the effects of interchanging motivational
variables does not rely upon the intensifying action of D but upon
the concept of habit-strength variation as a function of degree of
stimulus similarity.

Among the relatively few studies involving an interchange of
motivational variables, those by Miller (1948b) are of special
interest here because his interpretation of these phenomena is
identical with the habit-strength generalization hypothesis.

In Miller's first experiment on this problem, rats were taught
when thirsty to run down a narrow straight alley for water reward.
They were then divided into two groups: one group was made
hungry but was satiated with water; the other was satiated with
both food and water. When these subjects were placed in the
alley, the hungry but nonthirsty animals ran significantly faster
and actually drank more water at the goal than did those that
were neither hungry nor thirsty. Neither group, however, ran as
fast or drank as much as they had when thirsty.

The fact that the running and drinking responses did transfer
from a water-deprivation-food-satiation condition to a food-depriva-
tion-water-satiation condition, though with some decrement, is
explained by Miller in terms of generalization. The internal cues

122 THE MOTIVATION OF BEHAVIOR

present during original learning are similar to, but not identical
with, those present at the time of the tests for transfer. Transfer
should occur, therefore, but to an incomplete degree. These same
facts, of course, are perfectly consistent with the multiplicative-
drive hypothesis, since habits acquired under thirst should also
be evoked by the D due to hunger. As a matter of fact. Miller's
finding that hunger seemed to facilitate even the consummatory
response of drinking, which, on intuitive grounds, might be re-
garded as quite specific to thirst, fits the multiplicative theory very
nicely. It is a finding clearly corroborative of the notion that D
can catalyze any habit into action. Nevertheless, an associative
interpretation, based on repeated preexperimental associations be-
tween hunger stimuli and drinking, would also be defensible.

In Miller's second study, hungry rats were trained on an elevated
T maze to turn consistently to one side for food. After their learn-
ing had reached a high degree of proficiency they were thoroughly
satiated on wet mash in an effort to eliminate hunger completely.
They were then placed on the maze, where half of the animals in
the original group were given a one-second electric shock and the
others not. Since the shocked animals ran faster than the non-
shocked ones and made significantly fewer errors on each of sev-
eral successive runs, we may conclude with Miller that the shock
served as an effective substitute for the absent hunger. Miller's
associative view of this result is that the habit learned under
hunger generalized to pain. The internal cues characteristic of
these two states would seem to be so different, however, as to
make a generalization interpretation difficult. To overcome this
objection, Miller suggests that the generalization might be ". . .
based indirectly on stimuli produced by similar states of muscular
tension elicited by both drives, rather than upon common elements
directly present in the drives themselves" (p. 165). That is, the
animals made fewer errors when shocked because their vigorous
running produced proprioceptive stimuli much like those present
when they were hungry.

As a concomitant to the general problem we are considering,
we should also note that the presence of a need state such as thirst
can apparently increase the resistance to extinction of a response
learned under hunger. Both Webb (1949) and Brandauer (1953)

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 123

have shown that after hungry rats have been taught to perform
a simple instrumental response for food they will continue to
perform the response longer when they are thirsty and not hungry
than when neither thirsty nor hungry. Moreover, the more severe
the thirst, the greater the resistance to extinction. To explain these
findings an associative theorist would probably assume that as
thirst becomes more intense thirst stimuli become increasingly
similar to hunger stimuli. This assumption is defensible if one
considers only the dimension of stimulus intensity, but one might
also insist that as thirst grows stronger its characteristic internal
cues become qualitatively more distinctive and hence less similar
to those of extreme hunger. Grice and Davis (1957), who were
unable to confirm the findings of Webb and of Brandauer, have
suggested that when such results are obtained they may be due
to an increase in hunger induced by water deprivation.

Miller's theoretical treatment of the results of his T maze studies
was restricted to the use of the stimulus-generalization hypothesis
in explaining the degree to which one motivational variable could
be substituted for another. The same principles, however, may be
extended to the interpretation of behavior in complex choice situa-
tions following alterations in the strength of a single motivational
variable. Without going into the matter at great length, let us see
what might be predicted in the case in which a rat has learned
to turn left in a T maze under a moderate period of deprivation
and is then retested in the maze under either a long- or short-term
deprivation schedule.

The basic assumptions regarding the habits acquired in the
maze and the critical stimuli are shown in Fig. 4:6. If the left-turn-
ing reaction has been evoked more often than the right-turning
one, the habit strength of the former (H) will be stronger than
that of the latter {h). Both reactions, however, are conditioned
to the same stimulus compound consisting of internal stimuli of
moderate intensity (Sm) plus stimuli at the choice point of the
maze {Sep).

Imagine now that at the completion of the initial training the
deprivation period is lengthened (prior to a retest in the maze).
This should result in an increase in the intensity of the internal
stimuli. Conversely, if the deprivation period is shortened (partial

124 THE MOTIVATION OF BEHAVIOR

Fig. 4:6. Stimulus com-
ponents and relative habit
strengths presumed to be
characteristic of a situa-
tion in which an animal
has learned to turn left g

in a T maze for food un- -f- jj >./j

der moderate deprivation S^p
conditions. Sm = mod-
erately intense internal
stimuli; Sep = stimuli at
choice point of maze; H c

and h =z strong and weak -|- h — ^ jj

habits, respectively. S^p

right

satiation), the internal cues should become weaker. Presumably
these cues resemble one another in varying degrees, and they can
be arranged, therefore, on a continuum from weak {Sjw) through
medium {Sm) to strong (S/s). Such a continuum is represented
graphically as the stimulus dimension in Fig. 4:7. According to
Hull's postulates, the gradient of generalization for a strong habit
would correspond approximately to the upper curve in the figure
and the gradient for a weak habit to the lower curve. Since the
moderately strong stimuli were present at the time of training,
these cues would be more strongly associated with both left turn-
ing and right turning than would other intensities of internal
stimuli. From the hypothetical relations of Fig. 4:7, it follows
that the absolute strengths of both the right- and left-turning
habits should decrease regardless of whether the deprivation period
is increased (to produce Sis), or decreased (to produce Siw). More-
over, and this is the critical part of the analysis, a shift in the
length of the deprivation period in either direction from its training
value produces a decrease in the absolute difference between the
right and left habits. In the figure it will be seen that the difference
between H and h is larger than the difference between H' and h'
or H" and h". If the proportion of correct turns (to the left) is
determined by the absolute difference in the strengths of the two
habits, it follows that the efficiency of a rat's performance
in the maze should decline as the intensities of the internal cues
are shifted either up or down from an intermediate [training)
point. Although no one has apparently performed an experiment

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS

125

in exactly this way, Dinsmoor's (1952) study of discrimination
learning (see Chapter 3) is most nearly relevant. He found, it
may be remembered, that his animals did indeed perform more
poorly as deprivation level — actually, he used per cent body-weight
deficit, not hours of deprivation — was decreased from the value
used during original discrimination training. This fits the general-
ized habit-strength view quite well, though it is also consistent
with the multiplicative theory. But the performance of Dinsmoor's
subjects improved when hunger was made more severe than at the
time of initial training. This result does not follow from an un-
modified generalization theory such as the one of Fig. 4:7, which
predicts a decline in proficiency with lengthened deprivation. The
multiplicative-drive hypothesis, however, does predict improve-
ment. Apparently the generalized-habit view runs into the same
difficulties in attempting to account for behavior in a situation
involving two habits as it does in dealing with only one. Impair-
ment in performance with reduced deprivation can easily be en-
compassed, but improvements in performance following increased
deprivation cannot. The theory could doubtless be improved by

Fig. 4:7. Hypothetical generalization curves for a relatively strong habit (H)
and a weaker competing one (h). Weak, moderate, and strong internal
stimuli are plotted along the abscissa from left to right. It is assumed that
both reactions have been conditioned to the moderately strong internal
stimuli at the center of the dimension. With changes in the stimuh in either
direction, the strength of each habit decreases as well as the absolute differ-
ence between them.

126

THE MOTIVATION OF BEHAVIOR

the introduction of special assumptions, but considerable addi-
tional data would be needed before a judgment could be reached
as to its adequacy relative to a motivational theory.

Interpreting the Effects of Altered Motivational Variables in
Terms of Competing or Facilitating Reactive Tendencies. The
type of associative interpretation to be considered here rests on
the assumption that the manipulation of a motivational variable
produces behavioral changes through the evocation of habits that
can either facilitate or compete with the habit under investigation.
Fig. 4:8 indicates the stimulus elements, associative tendencies,
and responses to which this type of theory might appeal in explain-
ing performance decrements following a shift from long-term to
short-term deprivation. It has been assumed that the response is
one of learning to approach a goal for food under moderately
strong hunger. The upper half of the figure shows that during
original training the approaching response has become attached
to a stimulus compound consisting of Si, So, S3, S4, and Se. When
a shift is made to short-term-deprivation testing conditions, the
important elements are as shown in the lower half of the figure.

Internal cues due
to long-term -
deprivation

External cues

Training conditions

'1' -^a' ^3' "^4

-^S,

—H R

approach

Internal cues due
to short-term
deprivation

Satiation cues

External cues

Testing conditions
^8^,82 yh

approach

Fig. 4:8. Stimulus elements and associative tendencies assumed to charac-
terize a situation wherein an approach response has been learned under long-
term deprivation and tests are subsequently conducted under short-term de-
privation. These are the components of an associative interpretation that
relies upon competing habit strengths to explain the negative eflfects on per-
formance of shifts to short-term deprivation.

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS

127

For one thing, S3 and S4 are assumed to drop out under low mo-
tivation, which causes a weakening of the approaching habit
(shown by the lower case h). So far, this is identical with the
stimulus-deletion interpretation discussed in an earlier section of
this chapter. The new feature of interest at this point is the addi-
tion of stimuli S5 and Se- These may be identified as the typical
cues accompanying partial satiation, e.g., stimuli arising from a
distended stomach. Now it might be assumed that as a consequence
of past experience these cues have become associated with responses
of resting, lying down, or being inactive. If such is the case, then a
decline in performance under the satiation regimen could be
explained as a result of the introduction of the new tendencies to
rest which compete with the tendency to approach. Thus this inter-
pretation relies on the assumption that overt performance will be
weakened if tendencies to make incompatible reactions are simul-
taneously aroused.

The steps involved in applying this competitive-tendency view
to the case where the deprivation term is changed from short to
long are shown in Fig. 4:9. In the upper part of the figure, stimuli
S5 and Sq, with their already formed associated tendencies to rest

Internal cues due
to short-term
deprivation

Satiation cues

External cues

Internal cues due
to long-term
deprivation

External cues

Training conditions

■^81^,82-

Testing conditions

-H-

approach

'approach

Fig. 4:9. Elements and relations of an associative interpretation of perform-
ance facilitation following a shift from short- to long-term deprivation. The
presence of the tendency to rest during original learning is assumed to prevent
the development of a strong approach habit. Under long-term deprivation,
cues tending to elicit resting responses disappear and performance impro\'es.

128 THE MOTIVATION OF BEHAVIOR

are shown as present when the approaching response is being
learned to the complete stimulus complex. If during the course of
this learning the strength of the resting tendencies is never com-
pletely extinguished, the ultimate level of performance attainable
under short-term deprivation would be reduced by an amount
depending on the strength of the incompatible habits to rest.
Thus the net associative tendency to approach is represented by
a small h. Under long-term deprivation, stimuli S5 and Sq are
eliminated, and the subtractive (interfering) effect of the com-
petitive resting tendency (lower part of Fig. 4:9) disappears. As
a consequence, performance might improve, provided the new
stimuli S3 and S4 play an insignificant role, as they should since
they have not previously been associated with approaching. In
effect, then, performance facilitation may be explained by assum-
ing that an increase in deprivation time functions to remove or
eliminate reactive tendencies which have prevented performance
from reaching its potential maximum. It is also possible, however,
that if So and Se acquired some habit strength to elicit approach
behavior (dashed hne in upper part of Fig. 4:9) during the orig-
inal learning, their disappearance under long-term deprivation
would tend to impair performance and perhaps to offset any gains
due to the elimination of the resting responses. Strictly speaking,
however, this is a stimulus-subtraction mechanism rather than one
involving the interactions of habits.

Performance decrements following the substitution of one need
for another could be readily explained by the competitive-tendency
notion. In all such cases, the only assumption required is that the
new need produces new stimuli to which incompatible reaction
tendencies are attached. The competing-habit formulation can also
be extended to the results of studies (e.g., Fredenburg, 1956) in
which separate groups of subjects have been maintained under
different deprivation levels throughout the course of learning.
Perhaps interfering tendencies to rest become weaker as the
deprivation period is lengthened and asymptotic performance
level can therefore be elevated progressively.

When the competing-habit hypothesis is applied to behavior
in situations such as the T maze, the interpretation might be
structured somewhat as follows ; If performance declines, following

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 129

a shift from long- to short-term deprivation, it would be necessary
to suppose that competitive tendencies become stronger and inter-
fere more with correct than with incorrect habits. Conversely, if
training is carried out under a short-term deprivation schedule,
improvement following a lengthening of the deprivation period
would be ascribed either to the elimination of competing habits
or to the accrual of compatible ones.

As we have already noted, competing-habit interpretations of
the effects of variations in deprivation conditions have not been
formulated in detail. Clearly, the preceding suggestions toward
such a theory are largely ad hoc, and much remains to be done in
the way of refinement and extension. Especially needed are precise
statements as to how associative tendencies compete with or aug-
ment one another, and useful ways of defining competing habits
independently of the situations in which they are presumed to be
degrading performance. Incidentally, it is conceivable that the
most generally adequate associative theory would be one that
incorporates all of the best assumptions of the three versions
suggested here. Thus the notions of stimulus-element addition and
subtraction, of stimulus generalization, and of competing ten-
dencies might be combined into a single more potent theory of
the effects on behavior of changes in motivational variables.

Advocates of Associative Interpretations

At the start of our discussion of the various forms that an asso-
ciative interpretation might take it was noted that although no
one writer has presented us with a full-blown theory of this sort,
psychological literature contains numerous suggestions to the
effect that the construct of drive may be superfluous. Postman
(1953b), for instance, has clearly seen that since the manipulation
of certain so-called motivational variables may affect drive stimuli
as well as drive, it might be possible, and certainly more econom-
ical, to dispense with the concept of drive entirely and to base
one's interpretations of motivated behavior solely upon changes
in drive stimuli. Postman's (1953d) analysis of experiments on
motivational selectivity in perception is consistent with this sug-
gestion since, following the work of Solomon and Howes (1951),

130 THE MOTIVATION OF BEHAVIOR

he has placed heavy emphasis upon the contribution to "moti-
vated" perception of associative predispositions. It would be pre-
mature to attempt to identify Postman's views as characteristic
of any of the three associative versions presented above, but he
has appealed both to competing associative tendencies and to
modified stimuli in his interpretations of the effects on perception
of motivational variables.

Probably the most explicit associative theory of the effects of
motivational variables is that of Estes (1958), who has extended
his statistical theory of learning specifically to the problems of
motivation. This development, which appeared after the preceding
associative conceptions were written, and which can be presented
only in broad perspective because of its specialized terminolog)',
is essentially an extension and formalization of Guthrie's (1935)
contiguity theory. According to this view, the probability that a
response will occur depends upon the proportion of the stimulus
elements in a situation that have been conditioned to the response.
In the case of a variable such as food deprivation, the internal
drive stimuli assume a role of relatively great importance. An
increase in deprivation time has two principal consequences: (1)
the probability of occurrence of a drive stimulus increases, and ( 2 )
the probability of occurrence of stimuli characteristic of satiation
decreases. From these assumptions and from additional ones con-
cerning the relative weights, as behavior determinants, of these
cues, of conditioned cues, and of extraneous cues, Estes predicts
the outcomes of several experiments in which deprivation has been
manipulated. For example, the theory specifies that if training is
carried out at a single level of deprivation, asymptotic perform-
ance will be higher the more severe the deprivation. It is also
predicted that if a single group is trained at several deprivation
levels, as in Cotton's (1953) experiment (see Chapter 3 above),
asymptotic performance level should vary with deprivation if
behavior on all trials is counted, but not if those trials on which
competing responses occur are eliminated. The theory also predicts,
as was true of the associative hypotheses we have just considered,
that if training is carried out under a moderate level of deprivation,
performance should become worse if deprivation is suddenly en-
hanced. This expectation follows from the assumption that an

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 131

increase in deprivation will change the drive-stimulus complex by
introducing new elements not previously associated with the
response, and will thereby reduce the proportion of elements
conditioned to the response. Since this effect is attributed to the
tendency for the new elements to elicit interfering or competing
responses, this aspect of the theory is comparable to the third
associative interpretation presented above. But since the findings
of such studies as those of Deese and Carpenter (1951), Loess
(1952), and Hillman, Hunter, and Kimble (1953) contradict
this theoretical expectation, the drive-stimulus conception may
require modification. Bolles (1958) has also pointed out that the
theory may encounter difficulty in explaining (1) the correlation
of vigor of running with degree of deprivation when training has
been administered at a single level of deprivation, and (2) the
phenomena of incentive motivation. Nevertheless, Estes' formula-
tion, which he has also applied to the problems of need combina-
tion and stimulus generalization, undoubtedly constitutes the most
systematic attempt yet made to replace the concept of general
drive with purely associative mechanisms.

Bindra (1959), expanding on Estes' (1950) original theory, has
suggested that the interfering responses responsible for perform-
ance decrements following stimulus changes are actually "novelty
reactions" and that the performance decrement, therefore, is due
to, and directly dependent upon, the novelty of the test situation.
Although Bindra does not relate this hypothesis specifically to the
problem of motivational variables, it is apparent that it might be
extended to the effects of changes of needs or to changes in dep-
rivation severity. One wonders, however, whether it would be
meaningful to suppose that a rat would ever find his own internal
hunger or satiation stimuli to be novel even when they are con-
siderably different from what they were at some previous time.

Other hints that motivational phenomena can perhaps be inter-
preted without reference to a drive construct may be found in
the earlier writings of Hebb (1949), and in the theoretical inter-
pretations of Child and Waterhouse (1953), Meyer (1953), Mc-
Clelland, Atkinson, Clark, and Lowell (1953), Campbell and
Kraeling (1954), Farber (1955), Davis (1957), Holder, Marx,
Holder, and Collier (1957), and Meyer and Noble (1958).

132 THE MOTIVATION OF BEHAVIOR

Resume of Evidence Bearing
on Motivational Interpretations

Although associative conceptuaHzations appear to possess the
virtue of simphcity, it would be premature to conclude that the
idea of a general energizing drive must, therefore, be rejected
completely. One of the major reasons for retaining the hypothesis
of a general drive is that certain variables such as strong stimula-
tion and food or water deprivation have facilitative effects upon
surprisingly diverse kinds of responses. Considering intense stimuli
first, we find that electric shock or its aftereffects facilitates eat-
ing (Siegel and Brantley, 1951), drinking (Amsel and Maltzman,
1950), running speed and correct choices in a T maze after train-
ing under hunger (Miller, 1948ib), wheel turning to an auditory
signal (Nagaty, 1951), visual discrimination — when shock is ad-
ministered intracerebrally — (Fuster, 1958), reaction time (Johan-
son, 1922), human eyelid conditioning (Spence, Farber, and Tay-
lor, 1954), and of course, escape and avoidance behavior in many
different situations. Moderate degrees of muscular tension also
exhibit facihtative effects of considerable generahty. Courts (1942),
who has summarized these findings, states that all of the following
have been shown to be augmented by tension : conditioned salivary
responses in dogs, pursuit-rotor performance, the knee jerk, vibra-
tory sensitivity, reaction time, mental arithmetic, tapping, startle
responses, electric-shock thresholds, and ball throwing for accuracy.
In the case of food deprivation, it is interesting to note that in
addition to its familiar role as a facilitator of diverse instrumental
acts and of general activity (Siegel and Steinberg, 1949; Hall and
Hanford, 1954) it enhances reactions to intracranial electrical
stimulation (Brady, Boren, Conrad, and Sidman, 1957), drinking
(Miller, 1948b), escape from light (Bahrick, 1953), and the rein-
forcing effect of light onset (Davis, 1958). Moreover, as we have
previously observed, loud noises, bright lights, and air deprivation
also activate a variety of responses under particular conditions.

The important point in connection with all these observations
is that each of the several motivational variables has the capacity
to amplify activities whose manifest topographies and controlling
stimulus complexes differ extensively. In some cases it may be

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 133

reasonable to suppose that the internal (or other) cues produced
by the variable have been associated with the reference response
during previous periods of training. But it is difficult to imagine
that this has occurred in every instance where facilitation has been
observed. Wherever such associations cannot be invoked, a non-
specific energizing drive becomes a reasonable alternative.

The fact that motivational variables sometimes cause decre-
ments in performance must be noted, but it is not of crucial im-
portance for either the motivational or the associative theories.
Hull's multiplicative-drive theory interprets such findings by assert-
ing that drive is indeed increased and that either (1) associative
tendencies to make the wrong responses are initially dominant
over the correct ones, or (2) that the motivational variable pro-
vides new stimuli to which overriding competitive responses are
attached. The second of these is obviously the same conception
upon which associative interpretations rely.

A second major line of evidence favoring a general-drive theor}',
as against a purely associative view, is provided by studies in
which an increase in deprivation from that employed during train-
ing produces augmented performance. We have already reviewed
these studies and have observed that none of the associative ver-
sions, unless modified by special ad hoc assumptions, can readily
encompass this phenomenon. However, shifts from short to long
deprivation sometimes lead to the decrements in performance
predicted by associative theory (cf. Yamaguchi, 1952), and much
additional research is needed to determine the direction and extent
of the effects.

Other evidence consistent with the concept of general drive
stems from studies in which the combination of two different
motivational variables leads to better or more vigorous perform-
ance than does either alone. For instance, Amsel (1950(3) reports
that speed of running, as motivated by a conditioned pain-fear
reaction, is enhanced when a primary need for food is added to
the motivational complex. Similarly, Meryman (1952) has ob-
tained clear evidence that unconditioned startle responses in rats
are amplified more by a combination of hunger and conditioned
fear than by either separately. Jerome, Moody, Connor, and
Fernandez (1957) recorded the number of crossings made by rats

134 THE MOTIVATION OF BEHAVIOR

from one side to another of a multiple-door shuttle box, and noted
that the addition of hunger to an aversive light stimulus led to
more crossings than did either hunger or light alone. Probability of
error, however, proved to be unrelated to level of motivation. A
satisfactory associative explanation of these findings might be
developed, provided one could show that the motivation-variable
stimuli had been previously associated either with the reference
response or with responses capable of augmenting that response.
But this may prove to be difficult, especially in the Meryman
study, where one would have to suppose that the cues of hunger
and fear had somehow become conditioned to vigorous startle re-
sponses during the normal cage life of laboratory rats.

Experiments in which two appetitive variables are combined
have yielded conflicting results. Kendler (1945) found that the
addition of moderate periods (up to 12 hours) of water depriva-
tion to hunger led to an increase in the resistance to extinction of
a food-reinforced bar-pressing response, but that a reverse effect
was obtained when the period of water deprivation was increased
to 22% hours. Other studies, such as that of Verplanck and Hayes
(1953), indicate that consummatory activities either of eating or
drinking are reduced by the arousal of the opposite need; but
Powloski (1953) reports that discrimination learning is as effective
under both hunger and thirst as with either alone. In general,
the findings of these and related investigations do not support
Hull's view that two need states summate to yield a higher level of
general drive. The discovery of interactive effects among needs,
though not crucial to the concept of drive as a behavior determi-
nant, suggests the need for changes in the functions that have
been assumed to hold between drive and its deprivation an-
tecedents.

It is also worth noting, as Spence (1956, 1958) has shown, that
the specific hypothesis of a multiplicative relation between drive
and habit strength has been supported in a variety of studies,
specifically those in which performance curves obtained under dif-
ferent levels of deprivation or under different strengths of an un-
conditioned stimulus have been found to diverge as a function of
training. Similar curves obtained under different incentive levels
(i.e., magnitude of reward) also tend to diverge over training trials.

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 135

and since incentive is assumed to determine a motivational factor,
this too is consistent with the multipHcative interpretation.

In conclusion, it appears from the above summary of experi-
mental evidence that many diverse findings are consistent with
the concept of a nonspecific activating drive. Certain of these
facts, however, can also be explained as instances of the operation
of learned or unlearned associative predispositions. The principle
of parsimony supports the wisdom of exploring fully the utility of
associative interpretations, but until these theories have been
more fully developed it seems wise to retain both an intermediary
construct of general drive and associative mechanisms. In Chapter
9 additional evidence is presented from the field of physiological
psychology that is consistent with the general-drive conception.

Summary

In this chapter we have analyzed two general approaches to the
problem of how motivational variables function to modify be-
havior. One type of conception, termed a motivational interpreta-
tion, introduces, and places in a role of central importance, a
unique motivational construct such as drive. In Hull's multiplica-
tive-drive theory, which is used as the major example, drive is
described as a nondirective factor that is aroused by a variety of
antecedent conditions and becomes manifest in behavior through
its capacity to multiply existing associative tendencies (habits).
The multiplicative combination of drive with habits is assumed
to yield a quantity termed excitatory potential which, though
modified by other factors, is more or less directly reflected in be-
havior. Those who hold that intervening motivational constructs
are useful also maintain that motivational variables have stimulus
consequences in addition to their capacity to arouse drive and that
responses of many kinds are, or can become, associated with these
stimuli.

The principal implications of Hull's multiplicative theory are
reviewed and are seen to be the following: First, the fact that be-
havior is exhibited when, for example, an animal is hungry but
not when he is satiated is attributed to the multiplying effect of
drive and the resulting elevation of the appropriate excitatory tend-

136 THE MOTIVATION OF BEHAVIOR

encies to a suprathreshold level. Likewise, correlations between the
vigor of overt responses and time of deprivation are ascribed to
level of drive, provided that habit strength is held constant. Sec-
ond, if two or more reactive tendencies are present, the absolute
difference between their respective excitatory potentials will in-
crease as drive is strengthened and decrease as drive is diminished.
This leads to the prediction tli^t whenever the correct reactive
tendency is stronger than any pther, heightened drive will result
in improved performance. But if an incorrect tendency is strongest,
increased drive should lead to poorer performance. Conversely,
reduced drive can lead either to poorer or to better performance
depending upon which tendency is at the top of the response
hierarchy. These predictions will not be fulfilled, however, if the
motivational variable determining drive strength also serves as
the source of stimuli to which competitive or facilitative responses
have been associated. When the manipulation of a variable re-
sults in the introduction of salient cues, the hierarchical order of
habit strengths can be radically altered, and predictions based
upon the initial order must be modified appropriately. Third, if
in spatial- or discrimination-choice situations the habits corre-
sponding to the two responses are held equal, drive strength
should be unrelated to choice of the conect alternative.

The second major type of conception considered in this chapter
is one that rejects the need for constructs such as drive and ap-
peals only to variations in associative strengths in accounting for
the effects on behavior of changes in motivational variables. Be-
cause associative theories of this kind are just beginning to
appear, it has been necessar}' to be somewhat speculative concern-
ing the several forms that such theories may eventually assume.
One possible version of an associative theory might stress the
notion that performance is altered by a change in a motivational
variable from that used during training because some of the
stimulus elements associated with the response have been elimi-
nated and/or new elements have been added. One might thus
explain a decline in performance following a shift from hunger
to near-satiation; but without additional assumptions, performance
facilitation due to enhanced deprivation would be difficult to
interpret.

MOTIVATIONAL AND ASSOCIATIVE INTERPRETATIONS 137

A second variety of associative theory would emphasize the
idea that changes in motivational variables alter the intensity or
quality of cues and that alterations in existing associative strengths
would then occur in accordance with the principle of habit-
strength generalization. In effect, this principle is a guess as to how
the level of a reactive tendency is altered as a function of specific
modifications of the stimulus to which the response was originally
conditioned. In the extension of this principle to motivational phe-
nomena, emphasis is placed more upon the generalization of
internal stimuli than upon changes in external cues. This asso-
ciative model can encompass the effects of shifts from long to
short deprivation, changes from one need state to another, and,
with some additional assumptions, the transfer to a noxious stim-
ulus of a response learned under an appetitive need. As was true
of the first associative theory, however, the second encounters
difficulties in attempting to interpret performance enhancement
following changes from mild to severe deprivation.

A third associative interpretation might be developed which
stresses the possibility that new habits, either facilitative or com-
petitive, are brought into effective action by manipulations of
motivational variables. On this hypothesis, the original habit is
not changed to a major degree by the addition or subtraction of
stimulus elements or by variations in the quality or intensity of
the relevant stimulus complex. Instead, the motivational variable
modifies the effective strength of the original associative tendency
by evoking new habits and thus leads to different behavior. For
this interpretation to become widely useful, specific principles of
habit interaction would have to be developed and substantiated.

At the close of the chapter, evidence bearing on the concept of
a nonspecific drive was reviewed, and the tentative conclusion was
reached that although associative theories can successfully account
for a number of motivational phenomena, the construct of drive
is nevertheless supported by a wide variety of findings.

CHAPTER

Learned Responses as
Sources of Drive

As WE HAVE SEEN, particularly in Chapter 3, the various primary
motivational variables may influence behavior in profound and
widespread ways. In our society, however, with its high standard of
living, the large majority of individuals are seldom racked by the
pangs of severe hunger or by the tormenting sensations of intense
thirst, and the experience of intense or prolonged pain is the ex-
ception rather than the rule. It is, in part, because of this meliora-
tion of man's condition that the tendency to regard biological
needs and intense stimuli as the principal motivators of complex
human behavior has waned in recent times. In its place the wide-
spread conviction has arisen that the most significant source of
motivation for human beings lies in the elaborate experiential struc-
ture produced by learning and acculturation. Human beings are
said to be motivated to a significant degree by the so-called ac-
quired drives for success, prestige, security, power, affection, money,
property, and the like. As we shall indicate, it is questionable
whether specific directed tendencies of this sort should be called
"drives" at all, but it is clear that learning plays a major role in the

138

LEARNED RESPONSES AS SOURCES OF DRIVE 139

activities described by these terms. The infant human organism
certainly does not possess these tendencies, and it seems logical,
therefore, to assume that they represent complex relationships
among the diversified products of the learning process.

Before beginning our discussion of the acquired-drive problem,
a word or two concerning terminology is in order. In earlier chap-
ters it was pointed out that we are attempting to maintain a
clear distinction between a nondirective, nonspecific motivating
agency, e.g., Hull's D, on the one hand, and habits, with their
corresponding responses, on the other. As a consequence of this
attempt, we have avoided the use of the term "primary drives"
and have spoken instead of "primary sources of drive." The pri-
mary motivating variables, it may be recalled, were assumed to
affect behavior by contributing to drive, and/or by modifying
the stimulus complex {Smv) in certain ways.

This usage is continued in the present chapter and extended to
the traditional problem of acquired drives. To do so, however,
means that we shall seldom speak of acquired drives, in the
plural. If the word "drive" is used to mean a nonspecific, nondi-
rective, behavior-facilitating agency, then it is a construct to
which unitary properties should be ascribed. Many variables may
function motivationally, but since, in so doing, they are all affect-
ing behavior in the same way, it is simpler to refer to them as
"sources of drive," not as drives. This explains why the present
chapter has been entitled "Learned Responses as Sources of
Drive," rather than "Learnable Drives," or "Secondary Drives," or
"Acquired Drives."

The Acquired-drive Problem

That both learning and motivation are significantly involved in
secondary-drive phenomena is strongly implied by the words ac-
quired drive. For many psychologists the outcome of learning is
the acquisition of a response or a tendency to respond, and on
this assumption what is acquired in the case of an acquired drive
is a new or altered response. To have an acquired drive, therefore,
an organism must have learned to respond, and when this response \
is evoked the organism's behavior must be affected motivationally./

140

THE MOTIVATION OF BEHAVIOR

Thus the so-called acquired drives may be described as learned
responses having many of the general properties of all responses,
and, in addition, the capacity to affect other responses in the same
ways in which they are affected by motivational variables. Stated
somewhat differently, the acquired-drive problem demands answers
to the questions of how, and to what extent, the variables that
govern the learning of one response function as though they were
motivational variables when that response is interacting with
other responses. Within the framework of a motivational theory
such as Hull's, this idea is expressed by the assumption that some
actions are capable of adding increments to drive and thereby of
multiplying the habit strengths of other responses.

This conception of an acquired source of drive is represented in
the upper part of Fig. 5:1, where the learned (motivating) re-
sponse is shown as dependent upon learning variables and stimulus
variables (at the left), and as leading to an increment in general
drive (at the right). This drive increment, in turn, is assumed to
multiply the habit strength of a reference or indicant response.

Learning variables,
stimulus variables, -
etc.

Learned
motivating
response

Response-produced
->- increment to
general D

I /

Donse- /

Learning variables,
stimulus variables,
etc.

Response
produced

stimulus

Habit
strength

Reference
response

Fig. 5:1. Elements involved in two conceptualizations of an acquired source
of drive. The three components at the top of the diagram indicate that the
strength of a learned, motivating response depends primarily upon learning
variables and stimulus variables and that this response may bring about an
increment to general drive level. This drive increment may then affect a
reference response by multiplying the habit strength of that reaction. How-
ever, the motivating response may also affect the reference response by pro-
viding stimuli capable of modifying the latter's habit strength. If reliance is
placed upon the response-produced drive, the theory becomes associative-
motivational. But, if the entire explanatory burden is placed upon the re-
sponse-produced stimuli, the theory should perhaps be described as associa-
tive-associative.

LEARNED RESPONSES AS SOURCES OF DRIVE 141

Since both associative processes and the construct of general drive
are involved in this interpretation, it may be appropriate to term
it an associative-motivational conception.

Should one choose to omit drive entirely in dealing with the
so-called acquired motivations, appeal could be made to the
stimuli produced by the learned (motivating) response. These are
shown in the center of Fig, 5:1 and are indicated as affecting the
reference response by modifying its habit strength. Since the
strength of the learned motivating response rests upon associative
processes, and the mechanism of its influence upon the reference
activity is also associative, this second conception might be termed
associative-associative.

Thus for an associative-motivational theorist, learning variables
have motivational-variable consequences by virtue of their effects
upon the strength of a drive-producing response. But for the
associative-associative advocate, learning and stimulus variables
determine the original response and also mediate its effects,
through response-produced stimuli, upon indicant reactions. In
order to reduce our discussion to manageable proportions, we shall
proceed throughout the remainder of this chapter on the assump-
tion that an intermediary construct of drive is still desirable,
thereby adopting the associative-motivational position as a work-
ing hypothesis. There are good reasons for believing, however,
that the implications and ramifications of an associative-associative
theory are worth extensive exploration. The development of such
a theory would probably proceed along lines similar to those that
characterize the associative theories described in Chapter 4.

In order to simplify our exposition, the above description of the
associative-motivational view has been presented without mention
of the general drive produced by variables other than the learned
response. Within the theory, however, the strength of the motivat-
ing response and hence also the magnitude of the response-pro-
duced increment to D should depend upon other sources of drive.
This raises the interesting possibility that the response-produced
drive, when added to general D, can enhance the learned (motivat-
ing) response and hence itself. We shall make the simplifying
assumption, however, that this does not occur, the response-pro-
duced drive being assigned the power to affect other reactions,

142 THE MOTIVATION OF BEHAVIOR

but not itself. With respect to the latency of a motivating re-
sponse, this assumption appears quite reasonable, since the time
of onset of a given response can hardly be affected by a motiva-
tional increment that is contingent upon the appearance of that
same response. It is by no means certain, however, that a motivat-
ing response, once it has become fully aroused, cannot enhance
itself. An analogous self-augmenting process occurs in electronic
circuits under conditions where "positive feedback" from output
to input is possible, and the childhood temper tantrum may be a
psychological instance of the phenomenon. Moreover, if we take
the alleged motivating response of fear as an example, it would be
premature to insist either that fear reduction cannot function
to reinforce fear itself — perhaps this is why some fears are so
remarkably persistent — or that the presumed increase in drive ac-
companying fear cannot act to inhibit fear. Miller (1951) has
presented an illuminating discussion of certain aspects of these
problems, but it is the writer's belief that our knowledge of such
matters as the nature of reinforcement and the rise and fall of
motivating responses with time is too meager to warrant further
extensive speculation at present.

Although we have maintained that only learned responses or
readinesses to respond qualify as secondary sources of motivational
effects, this assumption is not as restrictive as one might imagine.
It does not mean that attitudes, opinions, expectancies, perceptual
readinesses, hopes, and the like, are necessarily eliminated from
consideration as acquired motivators. But it does mean that the
processes or events designated by these terms would have to be
treated, within the theory, as responses or readinesses to respond.
In many instances, this is not a difficult step. An attitude or an
expectancy is not a directly observable entity. Therefore, useful
inferences about expectancies or attitudes must rest ultimately
upon the fact — if and when it is a fact — that indicant responses of
such and such a kind have occurred in a specific environment.
When subjected to critical analysis, therefore, attitudes, hopes,
fears, opinions, and expectancies can probably all be reduced to
learned reaction readinesses. And to the degree that these re-
sponses affect other behaviors as do motivational variables, they

LEARNED RESPONSES AS SOURCES OF DRIVE 143

too would conform to our conception of acquired sources of drive.

The general way in which we have posed the acquired-drive
problem may seem abstract and impalpable because reactive tend-
encies and even many responses cannot be observed in any simple
and direct way. This need not be a matter for grave concern, how-
ever, since corresponding formulations of the problem can be
couched in the purely descriptive language of the empirical psy-
chologist or in the neurophysiological terms of the physiological
psychologist. Thus, the empirically minded scientist may wish to
deal only with directly observable relations between an organism's
molar behavior and other variables in its training or environment.
For such a psychologist the acquired-drive phenomenon would
be typified by the observation that one response is affected in cer-
tain specifiable (motivational) ways by the kinds and amounts of
practice, among other things, that an organism has had in making
some other response. In the case of the physiologically inclined
psychologist, the acquired-drive problem might be stated as a
search for the neurological or other bodily mechanisms involved
when one bit of behavior is affected in motivational ways by other,
previously learned, reactions. The manner in which the acquired-
drive problem is phrased will, of course, influence the kind of re-
search one is led to do, but the basic relations and critical questions
remain essentially unchanged throughout such transformations.

The foregoing analysis of the acquired-drive problem serves to
set the stage for discussions to follow, but it also raises numerous
questions for which satisfactory answers are not yet available. For
example, one wonders whether motivating responses can be dis-
tinguished from other responses by a consideration of such manifest
attributes as their latencies, durations, or anatomical bases. Or is
the course of the acquisition and the extinction of a motivating
response determined by the same variables (e.g., number of trials,
degree of distribution of practice, percentage of reinforcement)
that are now known to affect other reactions? Moreover, it would
be important to know something of the nature of the neural and/
or chemical mechanisms by means of which motivating responses
can influence other activities. And finally we shall eventually find
it desirable to determine precisely which of the criteria for the

144 THE MOTIVATION OF BEHAVIOR

identification of a motivational variable (cf. Chapter 2) have been
met on any given occasion when an acquired drive is alleged to
have been generated.

Conditioned Fear as a Source of Drive

Up to this point, our discussion of learned motivating reactions
has, of necessity, been rather general. To make our analysis more
specific we turn now to the consideration of one particular re-
sponse that seems to be rather well established as a motivator for
other responses. This is the response of conditioned fear, which,
in its motivating role, has usually been described as "the acquired
drive of fear or anxiety."

The general notion that a tendency to be fearful or anxious may
have motivating effects upon other responses, though historically
old, became significant for modern psychology through the writ-
ings of Cannon (1929) and Freud (1936). The translation of
this general conception into the more specific language of the
modern behavior scientist, however, was the work of Mowrer
(1939). According to Mowrer, anxiety or fear is a learned emo-
tional reaction to stimuli denoting the advent of a painful or
noxious event. It is a reaction acquired in accordance with the
associative laws of classical conditioning. Moreover, certain of its
behavioral effects are comparable to the effects produced by such
primary motivational variables as food deprivation and strong
stimuli. Specifically, fear sometimes seems to function as a gen-
eral energizer, and its reduction, following the elicitation of a new
directed response, may serve to reinforce the learning of the new
reaction. Since fear, described in this manner, is evidently learned,
and since its effects upon other bits of behavior resemble the
effects of motivational variables, it qualifies by our criteria as an
acquired source of drive.

Procedures Used in Conditioning Fear Reactions. Although psy-
chologists hold different views as to how fear is learned, all follow
essentially the same procedures in attempting to produce condi-
tioned fear in the laboratory. The training or conditioning trials
always involve paired presentations of a neutral stimulus — the so-
called conditioned stimulus {CS) — and a definitely painful or

LEARNED RESPONSES AS SOURCES OF DRIVE 145

noxious stimulus. The painful stimulus fulfills the role of the
familiar unconditioned stimulus {UCS). Quite frequently the
UCS is an electric shock intense enough to produce overt signs of
"emotional" excitement or disruption. In establishing condi-
tioned fear reactions in rats, for example, a buzzer, light, or click-
ing sound may be used as the CS. It is presented for perhaps a
second, though sometimes much longer, before the animal is
given one or more short, strong shocks. Typically, the paired ad-
ministration of the CS and UCS is said to result in the acquisition
by the CS of a tendency to evoke an anticipatory emotional (fear)
response that resembles, but need not be identical with, the pain-
produced reactions to the UCS.

When the responses in a conditioning experiment can be directly
observed and recorded, as in studies of eyeblinks or leg flexions,
strength of conditioning is usually estimated in one of two ways.
First, the CS may be presented alone, either following the comple-
tion of the conditioning trials, or at irregular intervals throughout
the course of conditioning. A conditioned response is defined as
the appearance on a CS-only test trial of a reaction of a certain
(arbitrarily defined) magnitude and/or latency. Second, on any
trial on which both the CS and the UCS are presented, a response
having the defined characteristics may appear between the onsets
of the CS and UCS, or after the onset of the UCS, but sooner
than if it had been a response to the UCS itself. Such responses
are tallied as conditioned responses.

These procedures are also occasionally used in studies of condi-
tioned fear. For example, if the response being recorded is a change
in heart rate or in skin resistance, the frequency or magnitude of
these changes to the CS alone or to the CS in anticipation of the
UCS, are commonly taken as indicating degree of conditioning. It
is rather difficult, however, to measure the presumed autonomic
components of fear in rats because of the animals' small size and
because of their tendency to struggle violently against restraint.
But since the laboratory rat is a convenient subject, investigators
desiring to use rats have found it necessary to develop other tech-
niques for estimating strength of conditioning in these subjects.
Specificalh , it has become common in recent years to estimate the
presence and strength of conditioned fear from observations of

146 THE MOTIVATION OF BEHAVIOR

the manner in which some other reaction of known strength is
modified by the presentation or by the termination of a CS that
has been paired with a noxious UCS. The response that serves as
the indicator may or may not be one that a naive observer would
describe as "fearful." Reactions such as eating, drinking, bar press-
ing, urinating, defecating, jumping, turning a wheel, and crossing
a hurdle have all been used as indicants of conditioned fear.

Studies of fear in which the more indirect methods are used
may be divided into three broad groups according to the ways in
which the CS is presented and the indicator response is affected.
The experiments in one group, which are consistent with the con-
cept of fear as an energizer, show that the reference response, if
elicited while the CS is on, tends to be enhanced or augmented.
Moreover, since degree of enhancement of the reference response,
which is itself never conditioned to the CS, tends to increase with
the number of CS-UCS pairings and to decrease during extinction
trials, it is reasonable to suppose that the changes in the indicant
reaction reflect strength of conditioning.

A second group of experiments consists of those in which, fol-
lowing fear-conditioning trials, tests are made to determine whether
the probability, speed, or amplitude of an indicant reaction in-
creases over a series of trials on each of which an ongoing CS is
terminated immediately after the response is evoked. When learn-
ing of the reference response is demonstrated under these condi-
tions, it is clear that the termination of the CS is reinforcing, and
this event thus meets one of our suggested criteria for identifying
motivational variables (cf. Chapter 2). And since the reinforcing
power of CS-offset varies with the number of CS-UCS pairings, an
associative variable, the conditions for an acquired source of drive
have been met. Usually the further assumption is made that the
CS arouses fear and that it is the reduction in fear due to the
termination of the CS that functions as the reinforcing agent.

In a third group of investigations, it has been found that the
presentation of a CS that has been paired with a noxious UCS
may inhibit or interfere with an ongoing overt response or response
sequence. By the proper choice of conditioning and testing meth-
ods it can be shown that the degree of interference suffered by the
indicant response increases as a function of the number of CS-

LEARNED RESPONSES AS SOURCES OF DRIVE 147

UCS pairings, and that the reference reaction tends to regain its
initial strength with repeated presentations of the CS alone as in
the customary extinction procedure. If the reference response that
is being weakened is followed by the onset of the CS, the condi-
tions and the resulting effect meet one of the criteria for the iden-
tification of motivational variables. And since this effect also
varies with associative variables governing acquisition and extinc-
tion, the picture is consistent with our general conception of an
acquired source of drive. Where the time relations of response to
stimulus onset differ from those mentioned here, however, the
inhibitory effects can perhaps be attributed to response-produced
stimuli and/or to CS-induced changes in the response hierarchy.
We shall return to this problem at a later point in this chapter.

We turn now to an examination of some typical experimental
studies from each of the three groups we have just described.

Studies of the Energizing Function of Fear. Although fear, at
times, may serve as a debilitating deterrent to action, ample evi-
dence from daily experience indicates that it often functions as a
potent goad to action. The reader has but to recall the frequency
with which frightened persons in stories, movies, and plays are
depicted as capable of almost superhuman feats of climbing, leap-
ing, and running, to realize that the notion of fear as an energizer
is exceptionally widespread.

Within the body of scientific writing, this basic notion has been
strongly supported by the physiologist Cannon (1929), whose
studies of hunger and thirst we have already mentioned. According
to Cannon, fear is an emergency reaction involving the mobiliza-
tion of energy resources of the body and their effective utilization
in expediting whatever responses are evoked by the situation at
hand. This conception has been widely adopted during the past
two decades, and it is only quite recently that the soundness of
certain of its elements has been questioned. Incidentally, Can-
non's views as to the dynamogenic properties of fear were not
based upon evidence from studies of behavior at the molar level,
but upon the biochemical and physiological consequences of fear.

Experimental studies by psychologists of the facilitating conse-
quences of fear are of relatively recent vintage. Relevant data,
though somewhat peripheral in origin, may be found in early

148 THE MOTIVATION OF BEHAVIOR

studies of conflict and especially in supporting experiments on
fear-motivated avoidance (cf. Bugelski and Miller, 1938; Brown,
1948) . In these studies hungry rats were first trained to run down a
short straight alley for food reward. They were then given electric
shocks of various intensities at the goal and were subsequently
tested without shock, when satiated, to determine how vigorously
they would avoid the region where they had been shocked. The
results indicated that the speed and vigor of the avoidance re-
sponses on test trials increased with strength of shock used during
training. Since shock was omitted on test trials, it may be presumed
that the avoidance was motivated by a learned tendency to be
fearful, which should vary with intensity of shock and with other
factors such as nearness to the point in the alley where the shocks
had been administered.

A direct attempt to determine whether an alleged fear-arousing
CS can intensify an overt skeletal response has been described by
Brown, Kalish, and Farber (1951). From the assumption that fear
as a learned source of drive should multiply existing reactive tend-
encies, and from clinical observations that anxious persons show
exaggerated startle responses, these investigators reasoned that
during the presentation of a CS that had been paired with shock,
rats should exhibit intensified startle reactions to loud sounds. Ac-
cordingly, an experimental procedure was arranged with the aims
of producing increasing fear during the course of fear-conditioning
trials (i.e., CS-UCS pairings) and decreasing fear during extinc-
tion trials, and of providing an opportunity for fear to show spon-
taneous recovery during rest periods following extinction periods.
Measurements were made at intervals throughout conditioning
and extinction to determine whether the amplitude of the startle
reaction varied concomitantly with the assumed variations in
fear.

The apparatus used in this study was a stabilimeter, by means
of which accurate graphic records of a rat's bodily jumps to a loud
sound could be obtained. A drawing of this device is reproduced
in Fig. 5:2. While confined in the rectangular box, the subjects
could be presented with visual or auditory stimuli and mildly
painful electric shocks. The fear-conditioning procedure involved
seven paired presentations of a CS (consisting of a buzzer and a

LEARNED RESPONSES AS SOURCES OF DRIVE

Confinement box

149

Polygraph

Fig. 5:2. Schematic drawing of stabilimeter used to measure the ampHtude
of rats' startle responses to loud sounds. {Adapted from Brown, Kalish, and
Farber,19Sl.)

light) and a UCS (electric shock) on each of four successive days.
Three test trials were also given per day, interspersed among the
seven fear-conditioning trials. On each test trial the CS was pre-
sented, but in place of shock a toy pistol was shot off to produce
a loud, sharp sound. The sound of this pistol almost invariably
elicited a definite startle response from the rats, even prior to
training of any kind. No shocks were ever presented on test trials.
The members of one group of experimental subjects were condi-
tioned and tested in this manner. The members of a control group
were tested in the same manner, but the temporal spacing of the
CS and UCS during their training trials was designed to minimize
or prevent the conditioning of fear to the CS.

The results obtained from tests interspersed among the condi-
tioning trials are summarized in Fig. 5:3. The curves shbw how
the animals of the two groups responded to the sound of the pistol
shot on successive days. The plotted points are medians of 15
values, each of which is a mean of three responses. The values
plotted at the zero point on the abscissa were derived from prelimi-

150

THE MOTIVATION OF BEHAVIOR
25

E
E
.E 20

a>
ir>
c
o

Q.
in

03
V

r 15

ro
"•a
<u

10
0

12 3 4

Successive days of training

Fig. 5:3. Median amplitude of startle responses of fearful and nonfearful rats
to a loud, sharp sound. The upper curve shows that experimental animals
presumed to be fearful jumped more and more vigorously to the sound as
the number of fear-conditioning trials increased. The responses of control
(nonfearful) subjects, however, did not change progressively or significantly
during the same period. {From Brown, Kalish, and Farber, 1951.)

nary measurements of startle-response amplitude on the day prior
to the start of training. These data were used to equate the mem-
bers of the two groups with respect to their inherent reactiveness
to the sound of the pistol.

An examination of Fig. 5:3 shows that the magnitude of the
startle responses evoked by the pistol shot in the presence of the
CS increased progressively during the course of conditioning for
the experimental animals but not for the controls.

The intervals between the CS and UCS and their orders of
presentation were chosen, of course, so as to maximize conditioned
fear in the experimental animals while minimizing it in the con-
trols. It seems reasonable to conclude, therefore, that the CS
elicited more fear in the experimentals than in the controls and
that fear functioned as a drive to augment unlearned startle re-
sponses. Stated in other terms, this portion of the experiment
provided evidence for an acquired source of drive, since an associa-
tive variable (number of conditioning trials) functioned as though

LEARNED RESPONSES AS SOURCES OF DRIVE

151

it were a motivational variable in its effects upon a reference
response which was itself not conditioned.

On three successive days following the period of fear condition-
ing, the animals of both groups were placed in the stabilimeter
and were given 12 nonreinforced test trials. On each trial the CS
was presented and the pistol was shot off, but the shock was
omitted. Since these trials would qualify as fear-extinction trials,
it was anticipated that fear would be weakened and startle-response
amplitude would decline. The curves for day 1 in Fig. 5:4 confirm
this expectation, since the startle responses of the experimental ani-
mals were relatively large at the beginning of the day, but dimin-
ished progressively with successive nonshock trials. By way of con-
trast, the control (nonf earful) rats started at a much lower level
and showed a much smaller decline during the shot-only trials. As
can be seen from the curves for day 2, the startle responses of the

12 3 4 5 6 7 8

Successive blocks of three nonreinforced trials

Fig. 5:4. These curves show the marked dedine in startle-response amphtude
exhibited by fearful (experimental) animals when a presumed fear-arousing
CS is repeatedly presented without shock. The smaller decrease in response
amplitude shown by the nonfearful control subjects appears to be the result
of adaptation. The rise of the solid curve at the start of further extinction
trials on day 2 may be due to the spontaneous recovery of fear in the interval
between the two tests. {From Brown, Kalish, and Farber, 1951.)

N. Day 1

Day 2

'^ \
\ V

• • Experimental group iV = 15

\
\
\

/
/

o — -o Control group A^=15

1 1 1 1

1 1

152 THE MOTIVATION OF BEHAVIOR

experimental rats showed a spontaneous increase in strength fol-
lowing a day's rest. This phenomenon is clearly consistent with
expectations from the laws of classical conditioning. If fear were
undergoing extinction during the 12 nonshock trials of day 1, we
should expect a period of rest to lead to some increase in its
strength. And if fear serves as a source of drive to enhance startle
reactions, a resurgence in the strength of the partially extinguished
fear should be reflected by a rise in startle-response amplitude, as
was indeed the case. Further support for the belief that fear is
weakened by trials without shock is provided by the progressive
fall of the solid curve during the nonreinforced trials of day 2.
Additional test trials given on day 3 showed no further systematic
effects, and they have been omitted from Fig. 5:4.

Another study supporting the view that conditioned fear may
serve as a learned source of drive is that conducted by Meryman
(1952). This experiment represents an important confirmation and
extension of the investigation of Brown, Kalish, and Farber.
Meryman was concerned with the problem of whether a primary
source of drive, such as food deprivation, would intensify sound-
induced startle responses in rats and whether hunger and condi-
tioned fear together would produce greater augmentation of startle
than either alone.

Meryman's apparatus was, in principle, the same as that used by
Brown, Kalish, and Farber. His rats were given fear-conditioning
trials and startle trials in a cylindrical, plastic-walled cage having
a grid floor. The cage was mounted on the end of a pivoted alu-
minum lever arm and any slight movements or startle responses of
the rats could be recorded on a polygraph.

Meryman's design called for four groups of animals. The animals
in group F-46 were motivated both by CS-aroused fear and by 46
hours of food deprivation at the time of their tests; group F-1 was
fearful but had been satiated one hour prior to the tests; group
NF-46 was not fearful but hadn't eaten for 46 hours; and group
NF-1 was not fearful and had been satiated one hour previously.

Following some initial adaptation trials to the sound of the pistol
shot alone, all animals were given three trials at 20-minute intervals
in the stabilimeter each day for ten days. On each of the first two
trials the animals were individually placed in the cage and the

LEARNED RESPONSES AS SOURCES OF DRIVE

153

amplitudes of their startle responses to the sound of the toy pistol
were recorded. On the third trial of each day, the procedure of the
first two trials was repeated for the fearful groups (F-46 and F-1),
save that a brief electric shock was substituted for the sound of the
shot. This was calculated to produce an association between fear
and the tactual, visual, and olfactory cues provided by the sta-
bilimeter cage. These environmental cues, rather than a buzzer or
light, served as the CS in this experiment. Thus the first two daily
trials provided measurements of startle responses as modified by
conditioned fear due to shocks received on the previous day or
days. The animals in the nonf earful groups (NF-46 and NF-1)
were also given two shot-only trials at the start of each day but
were not shocked when placed in the stabilimeter for their third
daily trial.

The results of Meryman's experiment are reproduced in Fig. 5:5.
The values plotted in this graph are means, based upon the two
daily startle responses to the shot for each of eight animals in each
group. Inspection of the curves reveals that on the first day, be-
fore any of the animals had had any shocks in the apparatus, all
four groups were about equal in their reactions to the shot. How-

B 16

« 8

Fearful -46 hours -
hungry

Nonfearful-1 hour hungry
9- — -9- 9 K?--

3 4 5 6 7 8 9

Successive blocks of two daily test trials

Fig. 5:5. Startle-response amplitude as a function of fear, no fear, intense
hunger, weak hunger, and their combinations. {From Meryman, 1952.)

154 THE MOTIVATION OF BEHAVIOR

ever, during the course of the ten days of training and testing, the
responses of the two shocked groups (F-46 and F-1) exhibited a
marked increase over their initial level. The responses of the fear-
ful-nonhungry group (F-1) showed a considerable increase in am-
plitude when compared with those of the nonfearful-nonhungry
group (NF-1), and the fearful-hungry animals (F-46) exhibited an
even more marked rise in their responses relative to the nonfearful-
hungry (NF-46) rats. These results clearly support our conception
of an acquired source of drive, since an associative variable {CS-
UCS presentations) affected the startle response as if a motiva-
tional variable had been introduced.

Two other findings by Meryman are of interest. First, the startle
responses of the NF-46 group were significantly greater than those
of the NF-1 group. This is important because it adds another re-
sponse to the list of those that are facilitated by food deprivation,
and because, to a degree, it allays the suspicion that startle, because
of some unique relationship to fear, can be facilitated only by
fear. Nevertheless, as the two middle curves show, fear, either be-
cause it is strong or because it is innately associated with startle, is
far more effective than intense hunger in enhancing startle. Sec-
ond, a comparison of the upper three curves reveals that fear and
hunger, when simultaneously present, are more potent than either
one alone. Precisely what the mechanism is by which these two
act in concert as facilitators is not clear. Conceivably, when hunger
and fear are combined the result is simply an increase in level of
general drive, multiplying the unlearned startle tendency. Or
hunger might function somewhat less directly by augmenting fear
and hence its capacity to affect startle.

In a subsequent study, Meryman (1953) has shown that the
amplitude of the galvanic skin response {GSR) to a weak auditory
click is enhanced by conditioned fear in human subjects. Fear
was presumably established by pairing a brief auditory cue (a hiss-
ing noise) with a mildly painful shock, an interval of 20 seconds
separating the two stimuli. Because of this long interval, almost
no conditioned GSRs were evoked by the CS even after a good
many trials. But on the occasional test trials, when a click replaced
the shock, the amplitude of the GSR to the test click showed a
marked and significant increase throughout the course of condi-

LEARNED RESPONSES AS SOURCES OF DRIVE 155

tioning. Since the click was never paired with shock, it cannot be
argued that the cHck was itself becoming a conditioned stimulus
for fear. Meryman concluded, therefore, that the noise had become
a conditioned signal for fear, in spite of the long interval between
noise and shock, and that fear intensified the unlearned tendency
to respond to the weak click.

In addition to these experiments in which the dynamogenic
effects of fear upon both startle and the GSR have been rather
clearly demonstrated, there are several other studies that provide
less direct support for the energizing conception of fear. Hunt
and Otis (1953), for example, report that defecation in rats, dur-
ing the presentation of a fear-arousing conditioned stimulus, in-
creases progressively during conditioning and declines during ex-
tinction. They interpret this finding to mean that the CS ac-
quired the power to "ehcit" defecation as one component of the
general emotional response. Alternatively, one might suppose that
this is not an associative phenomenon but a consequence of the
intensifying effect of fear-produced drive upon the defecation re-
sponse when it is evoked in the emotional situation. This view is
complicated, however, by the fact that another response, that of
bar pressing, which was also elicited in the situation, tended to
decrease as fear increased. Amsel (1950Z)) has also observed that
during a period when rats are recovering from the emotional after-
effects of shock, defecation starts at a high level and declines,
whereas drinking, which is initially inhibited, is gradually resumed.
Thus fear can apparently facilitate one response and inhibit an-
other at the same moment. Later on in this chapter we shall con-
sider the implications of this seeming paradox.

Some investigators have applied the term fear to the general
state of emotional excitement following the administration of elec-
tric shock, as well as to the conditioned emotionality aroused by a
CS in anticipation of impending trauma. A distinction should prob-
ably be made, however, between these two presumed types of emo-
tionality since their antecedent conditions, at least, are quite
unlike. In particular, it is evident that postshock emotionality
does not qualify as an acquired source of drive since a demonstra-
tion of its behavioral effects does not involve the manipulation of
associative variables. Residual emotionality appears, however, to

156 THE MOTIVATION OF BEHAVIOR

have drivelike energizing effects under certain conditions. Amsel
and Maltzman (1950), for example, found that the consummatory
response of drinking was enhanced following shock, provided the
shocks were administered in a different situation from the one
in which the rats were accustomed to drink. In a similarly designed
study, Siegel and Brantley (1951) observed that if hungry rats
were shocked in a separate place from where they were usually
fed, eating was facilitated. In both of these studies, the conclu-
sion was reached that postshock emotionality served as a source of
general drive to intensify whatever reaction was dominant in the
testing situation. Evidence contrary to this conclusion has been
obtained, however, by Kabrick and Farber (1952), who report that
postshock reaction times are lengthened in human subjects, and
by Brown, Meryman, and Marzocco (1956), who find that ampli-
tude of startle response in rats is diminished immediately after
shock. These negative findings may be due to the elicitation by
shock of competing responses, but further research is clearly
needed to determine the range of conditions under which facili-
tation and/or interference can be obtained.

Studies of the Reinforcing Effect of Fear Reduction. All of the
experiments in this group show that when a CS has been paired
with a noxious UCS, terminating the CS after the evocation of a
new response serves to reinforce the learning of that response.
Since the capacity to function as a reinforcer is one of our sug-
gested criteria for identifying motivational variables, and since
this capacity varies in these studies with CS-UCS pairings (an
associative variable) the operation of an acquired source of drive
is strongly indicated.

The best-known experiment of this group (Miller, 1948cz) has
been quite influential in establishing the view that fear functions
as a learned source of drive. The basic apparatus used by Miller
was a rectangular box divided by a vertically sliding door into two
separate compartments. One of these was painted white and had
a grid floor, the other was black with a smooth floor. Twenty-five
rats were given a series of preliminary training trials on which they
were taught to escape electric shock in the white side by running
through the open door into the black compartment. This training
was also designed to produce a strong association between fear and

LEARNED RESPONSES AS SOURCES OF DRIVE 157

the visual and tactual cues of the white box. During subsequent
learning trials, the door was closed, but it would open automati-
cally if a rat rotated a small wheel located directly above it. To
determine whether reduction of the fear conditioned to the white
box {CS) during the shock-escape trials would actually be rein-
forcing, Miller placed his rats in the white compartment without
shock. Since the door was closed, they were faced with the task
of learning to rotate the wheel in order to be released from the
white box. The conditions were such that if an animal made the
new wheel-turning response within 100 seconds, the door dropped,
and the fear aroused by the white-box cues could be reduced by
escaping into the black section. If the wheel-turning response was
not made within this period, the animal was removed from the
apparatus to await another trial.

The results obtained on the nonshock trials showed that if a rat
succeeded in turning the wheel a number of times during the early
part of the series, the new response was rapidly learned. According
to Miller, escape from the fear-eliciting white-box cues served to
reinforce wheel turning and to bring about a marked increase in
the speed with which the act occurred. Actually 12 of the 25 rats
did not turn the wheel often enough at first, and with repeated
nonshock trials in the white side, their conditioned fear was extin-
guished before they could learn the new response. The 13 rats
that did learn, however, also learned a second new response with-
out any further shock trials. After wheel turning had been learned,
the door was closed and the controlling circuits arranged so that a
small metal rod had to be depressed to operate the door-release
mechanism. During further trials with this arrangement, the rats
gave up their earlier response of wheel turning, since it was no
longer effective in leading to fear reduction, and acquired the
response of depressing the bar. Shock was never administered on
the learning trials, and Miller concluded, therefore, that fear was
aroused by the white-box cues and that the elimination of those
cues following escape into the black compartment led to a reduc-
tion of fear, which reinforced wheel turning and bar pressing.

Using Miller's study as their point of departure, Brown and
Jacobs (1949) performed an experiment which provided further
support for Miller's interpretation. From an analysis of Miller's

158 THE MOTIVATION OF BEHAVIOR

study, these experimenters reasoned that some other kind of rein-
forcement might have been operating in addition to, or instead of,
fear reduction. Specifically, during the initial training phase of
Miller's experiment, the animals had received a number of shock-
motivated trials in escaping from the white box to the black box.
Quite conceivably, therefore, during the subsequent learning trials
the animals may have been frustrated when prevented by the
closed door from further escape. If so, the anger or frustration
produced by interference with escape could have served as a drive,
and its reduction, following wheel turning and escape, could have
been reinforcing.

Following this line of thought, Brown and Jacobs designed their
study so that no escape responses were permitted or rewarded dur-
ing the initial fear-conditioning trials. These trials were ad-
ministered alternately in each section of a black-walled, two-com-
partment box. Each fear-conditioning trial consisted of the paired
presentation of a compound CS (interrupted tone plus blinking
light) and a pulsating electric shock (UCS). The animals could
not jump from one side of the box to the other on these trials, and
the durations of the CS and the UCS, as well as the interval be-
tween them, were completely independent of the animals' activi-
ties. In all, 22 trials of this kind were given, 10 on each of the
first two days, and 2 at the start of the third day. Half of the trials
were given in each side of the box so that fear, if it became con-
ditioned to the cues provided by the apparatus, would be equal on
both sides.

Tests for the reinforcing effects of fear reduction were conducted
by putting the animals into one side of the box and turning on the
compound CS. At the same time, a guillotine door in the partition
separating the two compartments was lifted and the animals were
permitted to jump over a low hurdle into the opposite compart-
ment. If they made this new response within 60 seconds, the door
was closed behind them and the CS was turned off. No shocks
were ever administered on these trials. It was anticipated that if
the CS aroused fear, its cessation following the hurdle-jumping re-
sponse would be rewarding and the performance of that response
would improve. Animals not shocked during the initial trials
served as controls.

LEARNED RESPONSES AS SOURCES OF DRIVE

159

The time taken by an animal to jump over the hurdle after the
door was opened provided the necessary performance data. In-
dividual trial latencies were transformed into logarithms to nor-
mahze the distributions for statistical analyses, and the resulting
means were plotted as in Fig. 5:6. In this figure the solid line
shows that the hurdle-jumping latencies of the experimental (fear-
ful) rats decreased progressively during the first 20 trials of hurdle
jumping. The significant drop in these latencies suggests that the
barrier-crossing responses of these rats were being reinforced by the
cessation of the CS. The performance of the control-group animals
contrasts sharply with that of the experimental rats. Whereas both
groups crossed the hurdle with about the same latencies at first,
the controls showed no tendency to improve their level of perform-
ance with additional trials. In fact, a statistical comparison of the
first and last points of the dashed-line curve indicated that the
performance of the control animals became significantly worse
over the 40 trials. It was concluded from this study, therefore, that/
fear was developed in the experimental animals during the initial

2.5

2.0-

£ 1.5

1.0
0

<^--'

^ ■ —

--'"(

^Control group

, Experimental group

1 . .

1 I 1

3 4 5 6

Blocks of five trials

Fig. 5:6. Hurdle-jumping latencies for fearful and nonfearful rats (N = 8
per group) as a function of trials. The animals of the fearful (experimental)
group show a marked decrease in latency (improvement in performance)
during the first 20 trials, whereas the controls do not. The only reinforce-
ment was that provided by the cessation of fear-arousing cues following the
response. {From Brown and Jacobs, J 949.)

160 THE MOTIVATION OF BEHAVIOR

conditioning trials even though escape responses were neither
permitted nor reinforced, and that the reduction of fear, following
the cessation of the CS when the response occurred, functioned to
strengthen the tendency to perform that response.

Kahsh (1954), using procedures patterned after those of the
Brown-Jacobs experiment, has carried out the most complete study
of the effects on fear of varying numbers of acquisition and ex-
tinction trials. Previous experiments by Miller and Lawrence
(1950) and by Gwinn (1951) had yielded somewhat obscure data
concerning fear as a function of the number of shock-reinforced
trials.

In his study Kalish used four principal groups of animals that
were given 1, 3, 9, and 27 fear-acquisition trials, respectively, in a
small, grid-floored, gray box. Each trial involved the presentation
of a buzzer plus a blinking light {CS) for five seconds, with a one-
second pulsating shock (UCS) being turned on during the last
second of the CS. As in the Brown-Jacobs experiment, these ac-
quisition trials were uncomplicated by the learning of escape or
avoidance responses. Each of the four basic groups was divided
into four subgroups. These were given 0, 3, 9, and 27 extinction
trials {CS presented alone), respectively, following the initial
conditioning trials. Immediately after extinction, the 16 groups
were given a chance to learn a hurdle-jumping response with the
CS-cessation providing the only reinforcement. Because of these
procedures, the 16 groups should have differed, at the time the
hurdle-jumping trials began, with respect to the amount of fear
remaining after varying numbers of fear-acquisition and fear-extinc-
tion trials. Differences in proficiency in learning the hurdle-jump-
ing response would be expected to reflect these differences in
residual fear.

When the hurdle-jumping latencies for the 16 groups are com-
bined according to the original number of fear-conditioning trials,
irrespective of number of extinction trials, and are plotted against
the trials, the four curves of Fig. 5:7 are obtained. Here we see that
the group given the maximum number of fear-conditioning trials,
i.e., 27, shows the greatest decrease in latency as a consequence of
the reinforcement provided by the termination of the CS. As the
number of conditioning trials decreases to 9, 3, and 1, the degree

LEARNED RESPONSES AS SOURCES OF DRIVE

161

1 2 3

Blocks of three learning trials

Fig. 5:7. Hurdle-jumping latencies on successive blocks of three trials during
which the response was reinforced by the cessation of an alleged fear-arousing
CS. Each curve was obtained by pooling the data from four subgroups of
eight animals given differing numbers of extinction trials. {From Kalish,
19S4.)

of learning in the hurdle-jumping tests declines, though some
learning seems to have resulted from only a single fear-condition-
ing trial. Thus fear seems to be acquired very quickly, though its
strength continues to rise with additional reinforcements up to at
least 27 trials.

When the data from Kalish's experiment are regrouped to show
the effects of variations in number of fear-extinction trials, the
curves of Fig. 5:8 are obtained. Here again the findings accord
well with expectation, since animals given no extinction trials
learn most rapidly, and learning becomes progressively worse as
the number of extinction trials is increased.

The procedures and findings of Kalish's study, it may be seen,
are consistent with the concept of acquired sources of drive devel-
oped at the start of this chapter. Thus, number of acquisition
trials and number of extinction trials, though clearly associative
variables, affect the hurdle-jumping reference response, which was
not elicited while these variables were being manipulated, much as
we might expect it to be affected by a motivational variable.

Concerning data such as these, however, it must be remembered
that although decreases in response latency provide evidence for
the presence of fear, and differences among such decreases indi-

162

THE MOTIVATION OF BEHAVIOR

1 2 3

Blocks of three learning trials

Fig. 5:8. Hurdle-jumping latencies for groups differing with respect to number
of fear-extinction trials. The data from four subgroups of eight animals each,
given different numbers of fear-acquisition trials, were combined to form each
of these four curves. {From Kalish, 1954.)

cate differences in degree of fear, absolute reaction latencies do not
reflect strength of fear in any direct manner. This is because the
hurdle-jumping training is carried out in the absence of further
shock trials. During such trials fear must be presumed to be under-
going extinction. Thus proficiency in the performance of the new
response may be increasing, but this increase is occurring while
the absolute strength of fear is doubtless declining.

The experiments described in this section on the reinforcing
effects of fear reduction may be taken as representative of typical
research in the area. Other studies by Miller and Lawrence (1950),
Gwinn (1951), and Mowrer and Lamoreaux (1942) have also
provided evidence that fear reduction is reinforcing.

Studies of the Inhibitory Effects of Fear. We have seen that if a
reference response is elicited while a CS for fear is on, enhance-
ment of the response may be observed. Under other conditions,
however, especially when the CS is presented after a response (or
response chain ) has been initiated, activity may be inhibited rather
than augmented. At first sight this negative effect seems to conflict
with the notion of fear as a motivator. Nevertheless, it can be in-
terpreted as an instance of an acquired source of drive, if one
grants that the inhibition or abandonment of responses that are
followed by the introduction of a motivational variable is a legiti-

LEARNED RESPONSES AS SOURCES OF DRIVE 163

mate criterion (Chapter 2) for the identification of such variables.
On this basis, evidence for an acquired source of drive would be
provided by the observation that as the number of fear-condition-
ing trials (associative variable) increases, the likelihood that re-
sponses which are followed by the CS will be inhibited (motiva-
tional variable) should also increase.

One of the first experimental demonstrations of inhibiting ef-
fects of this kind has been reported by Estes and Skinner (1941).
In their experiment rats were trained initially to press a bar to
obtain a pellet of food. By presenting pellets at fixed intervals of
time rather than after every response, the rats were taught to re-
spond at a continuous rate. That is, they would depress the bar
repeatedly even though only one pellet was given them every
four minutes. Subsequently, fear reactions were conditioned (while
the animals were pressing the bar) by turning on a tone [CS) for
several minutes and giving the rats a strong shock at the time the
tone was terminated. At first the rats continued to respond at
their usual rate while the tone was sounding, but as the number
of tone-shock trials increased, the onset of the tone produced a
more immediate and more nearly complete cessation of bar press-
ing. Animals tested under low hunger drive did not exhibit this
effect to as marked a degree as did others tested under a strong
hunger drive. This latter finding is of special interest since it is
consistent with the view that fear may be intensified by the pres-
ence of hunger. This possibility was suggested above in connection
with Meryman's startle-response study.

The Estes-Skinner technique for investigating the behavioral
consequences of fear has proved to be rather generally useful. Hunt
and Brady (1951 ) and their associates, in particular, have employed
the method repeatedly in their studies of the effects of electrocon-
vulsive shock (ECS) upon fear. Though space does not permit a
detailed review of their experiments here, these investigators have
demonstrated rather clearly that the immediate aftereffects of
ECS treatments are to abolish conditioned fear. Rats given ECS
"therapy" are not deterred from repetitive bar pressing when a
stimulus is presented that was formerly followed by strong shock.

Amsel (1950fo) has shown that conditioned fear inhibits the
consummatory response of drinking in thirsty rats, provided fear

164 THE MOTIVATION OF BEHAVIOR

and the consummatory act have been associated with the same
cues. In his experiment rats were given prehminary training in
running from the larger of two similar white compartments into
the smaller to drink from a calibrated water container. During this
training, which was carried on for 24 days under a constant water-
deprivation regimen, the amount of water consumed during a five-
minute period on successive trials increased in a smooth, negatively
accelerated manner until a stable level was reached. At this point
the animals in one group were placed in the larger box and were
given two shocks per day for four days. Control animals were
also put into the larger section but were not shocked. Subse-
quently, when the animals of both groups were again permitted
to run to water, those in the shocked group showed a sharp drop
in water consumption. The water intake of the no-shock control
group, however, did not change following the four-day no-drinking
period. During additional no-shock test periods, the shocked ani-
mals exhibited a gradual increase in the amount of water drunk
during the five-minute period. In interpreting his findings, Amsel
assumed that fear was conditioned (in the experimental group) to
the cues of the larger white box and generalized to the smaller
white compartment. The decrease in water intake of these animals
on the postshock trials was ascribed to the presence of "anxiety-
motivated" competing responses. Thus fear did not facilitate
drinking because its presence led to the appearance of new re-
sponses which interfered with drinking. One of these competing
responses may have been emotional defecation, since, as we have
previously noted, the frequency of defecation by the fearful ani-
mals was extremely high at the time when water consumption was
low. Moreover, defecation decreased as fear was extinguished and
water consumption rose.

This concludes our discussion of selected studies illustrating the
inhibitory effects of fear upon other responses. Unfortunately, we
are not yet in possession of a large body of experimental data in
this area, and interpretations and generalizations must be cau-
tiously proposed. We need, for example, more detailed information
concerning the kinds of responses that are inhibited by fear, in-
formation as to the strengths of those responses relative to others.

LEARNED RESPONSES AS SOURCES OF DRIVE 165

and information about the specific characteristics of those fear-
conditioning situations that lead to maximum inhibition.

Fear as Both Inhibitor and Energizer. At the start of the preced-
ing section it was noted that an apparent contradiction is imphcit
in the fact that fear sometimes functions as an energizer and some-
times as an inhibitor. However, it was asserted that within the
framework of the present conception of acquired sources of drive
these opposite efFects are not paradoxical, since one of our criteria
of a motivational variable is that its introduction following a re-
sponse tends to lead to the abandonment or inhibition of that re-
sponse. But this interpretation is applicable only when one can
provide reasonable evidence that the inhibited response was in-
deed elicited before the onset of the fear-arousing CS. Unfortu-
nately, this is usually not easy to do, especially in instrumental con-
ditioning situations where the onset and/or termination of the
CS may not be precisely controlled. The evaluation of this concep-
tion must be delayed, therefore, pending the accumulation of addi-
tional data.

An alternate hypothesis as to why fear has inhibitory properties
can be developed along purely associative lines. Let us suppose, for
example, that bar pressing is the reference response. If that re-
sponse occurs consistently prior to fear conditioning, then the bar-
pressing habit must be stronger than other habits. Moreover, since
bar pressing is not enhanced by the drive increment that might be
produced by the first fear-conditioning trial, but is either slightly
depressed or is unchanged, it must be further postulated that one
trial is sufficient to shift some habit other than bar pressing into a
position of dominance. Thus a tendency to freeze or to crouch or
to move away from the region of the bar could have become
slightly stronger than the tendency to press the bar. Whatever the
nature of the newly dominant tendency, it must be regarded as
one that can prevent or interfere with bar pressing. It could be
either an innate or a learned response to fear-produced internal
stimuli. As fear increases with training, the internal cues would be
expected to increase in intensity, or more of them might rise above
a threshold value, producing an increase in the superiority of the
interfering tendency relative to the bar-pressing tendency.

166 THE MOTIVATION OF BEHAVIOR

Should one choose to do so, the construct of D could be added
to this associative theoty without changing its predictions. Both
the new competing habit and the bar-pressing habit would be mul-
tiplied by the drive due to hunger and by the mounting fear-pro-
duced drive. Crouching or avoidance reactions and bar pressing
should therefore both become stronger, but with the increase in
drive and in the difference between the two habits, the strength
of the interfering excitatory tendency would become progressively
greater relative to the bar-pressing excitator)^ tendency. Conse-
quently, overt bar pressing should decline, even though the ab-
solute magnitude of its excitatory tendency was rising. As we have
already seen, the multiplicative-drive assumption alone cannot ex-
plain declining performance under circumstances in which drive
level is presumably increasing. One must further hypothesize that
the variable leading to increased D also serves to bring a new, in-
compatible reactive tendency into a position of ascendance over
the old. This is clearly an associative assumption, but its use per-
mits one to retain the hypothesis of D as a multiplying agent even
where an indicant reaction is becoming overtly weaker.

Variables Influencing the Strength of Fear. Some of the vari-
ables determining strength of fear have already been mentioned in
our discussions of specific experimental studies of fear. There are
numerous others, however, which, though they cannot be treated
here in detail, are of sufficient importance to mention. For an ex-
cellent analysis of many of these the reader is referred to Miller's
(1951) treatment of learnable drives and rewards.

The most general statement one can make concerning the
strength of fear is that it is influenced by many of the same associ-
ative variables that determine the strength of other conditioned
responses. For example, fear tends to become stronger as the in-
tensity of the noxious UCS is increased (Miller and Lawrence,
1950). And if electric shock of near tetanizing intensity is used,
with dogs at least, fear becomes so firmly ingrained that its ex-
tinction is accomplished only with great difficulty (Solomon, Ka-
min, and Wynne, 1953). Fear also tends to increase progressively
as the number of paired presentations of the CS and UCS is in-
creased (Brown, Kalish, and Farber, 1951; Kahsh, 1954). Beyond a

LEARNED RESPONSES AS SOURCES OF DRIVE 167

certain point, however, fear may decline with further conditioning
trials. This possibility is suggested by Libby's (1951) finding that
80 fear-conditioning trials appeared to yield weaker fear than 40.
Electric shocks of long duration may also produce stronger fear
than short shocks, but the evidence is far from convincing
(Mowrer and Solomon, 1954). Fear also is known to vary with
the length of the interval between the CS and UCS (Libby, 1951;
Murfin, 1954), and perhaps with degree of massing of fear-condi-
tioning trials (Armus, 1954). Moreover, fear reactions may be
more intense when an organism is hungry than when it is not
(Estes and Skinner, 1941; Meryman, 1952).

Although existing studies of factors influencing strength of fear
are illuminating, systematically gathered data in this area are still
relatively meager. Much additional work certainly remains to be
done before we can be reasonably certain as to what the optimum
conditions are for the acquisition of fear. Investigations are rare, for
example, of the generalization of fear or of the effects on fear
of shock duration and changes in the characteristics of the CS.
More data are also needed on fear as a function of the distribution
of both acquisition and extinction trials.

Concerning the problem of the elimination or extinction of fear,
it seems likely that any factor tending to increase strength of fear
should also enhance its resistance to counteracting influences. As
is well known, degree of resistance to extinction is often used as
a measure of strength of fear and other responses. The procedures
for the conditioning of fear are those of classical Pavlovian condi-
tioning, and experimental extinction, therefore, is the basic method
for eliminating fear. Thus, to weaken fear it is only necessary, in
principle, to present the fear-arousing CS without the UCS. We
have already seen from Kalish's study (1954) that fear declines
progressively with the number of such nonreinforced presentations
of the CS. Solomon and Wynne (1953) have suggested that fears
often show remarkable resistance to extinction because the individ-
ual's avoidance responses remove him too quickly from the pres-
ence of the fear-arousing CS. Presumably, for extinction to be suc-
cessful, fear must be more or less fully aroused by the CS and yet
not be reinforced by the UCS. These authors have also hy-

168 THE MOTIVATION OF BEHAVIOR

pothesized that traumatically strong shocks may produce condi-
tioned fears that can never be completely eradicated by normal
extinction procedures.

The process of fear extinction can be accelerated by massing
extinction trials (Burros, 1949) or by permitting or persuading the
subject to engage in appetitive activity while in the presence of the
CS. This latter relationship has been demonstrated experimentally
by Farber (1948). Those of his rats that ate in the fear-arousing
environment lost their fears more quickly than animals that spent
the same amount of time in the same environment but did not
eat. In Jones's (1924) classic demonstration of this principle, a
child's fear of a rabbit was overcome by first presenting the rabbit
at a distance while the child was eating and by then bringing the
animal closer and closer to the child on successive occasions.

Conditioned fear in rats can also be very nearly eliminated by
the use of electroconvulsive shocks (Hunt and Brady, 1951), as we
have already noted, and by the elicitation of audiogenic seizures
(Brady, Stebbins, and Galambos, 1953). Moreover, alcohol, which
is well known for its ability to reduce human fears and anxieties,
tends also to reduce the fears of rats in laboratory conflict situa-
tions (Conger, 1951).

The problem of precisely which variables govern the learning of
emotional reactions such as fear has engaged the attention of
numerous theorists. For some, the learning of both emotional and
nonemotional responses is governed by the same principles; others,
however, hold that the known facts cannot be encompassed unless
two or more different laws or principles are invoked. Marshalling
and analyzing the evidence relative to this problem is a fairly
lengthy matter, and we postpone it, therefore, to our final chapter
which is devoted to theoretical issues of this sort.

Anxieties or Fears as Learned Sources
of Drive in Human Behavior

Thus far our discussion of acquired-drive problems has been
restricted, for the most part, to laboratory studies of, and theories
derived from, conditioned fear responses in animals. We have
followed this course because the most carefully controlled experi-

LEARNED RESPONSES AS SOURCES OF DRIVE 169

ments have been done with animals, not because the study of their
artificially induced fears is, in itself, to be regarded as of primary
importance. The study of fear in animals is, of course, a perfectly
legitimate area for scientists to explore. But most research workers
wish to know whether functional relations obtained with one or-
ganism can be generalized to other organisms, including man.
Actually, the process is a two-way one. We can learn important
things about rats' fears from human beings, and vice versa. From
everyday experiences with our fellow men and from laboratory
studies, we often get hunches as to the ways in which certain vari-
ables affect behavior. But in testing out these hunches, we may find
it desirable to use animals as subjects. This is often true when we
desire to minimize the role of verbal mechanisms or when it is
necessary to employ fear-arousing unconditioned stimuli of a
traumatically intense nature. Studies such as those of Solomon,
Kamin, and Wynne (1953) on traumatic-avoidance learning in
dogs could never have been done with human subjects.

That human fears, anxieties, and feelings of insecurity have
widespread effects upon behavior is almost universally acknowl-
edged. In Freudian (1936) theory as in others, feelings of anxiety,
guilt, and the like, play a major role in the interpretation of nor-
mal behavior. These feelings or reactions, moreover, are believed
to constitute the cornerstones of the bizarre personality structures
of the neurotic and the psychotic.

Human Fears as Learned Responses. Although tendencies to
become anxious could conceivably be transmitted through genetic
channels, human beings, by and large, must learn to be anxious.
During a child's early years there are innumerable opportunities
for it to acquire fears of certain stimulus objects in certain situa-
tions. We know little, unfortunately, about how these anxieties
are learned, but it seems reasonable to believe that they are ac-
quired in much the same ways as are conditioned fears in animals.
For instance, the sight of a flaming match may serve as a CS,
with the subsequently experienced burn providing the UCS. Since
similar sets of events occur repeatedly, ample opportunities are
provided for children to acquire fears of whatever objects society
regards as dangerous. Indeed, in order to save children from injury,
parents spend a great deal of time making certain that youngsters

170 THE MOTIVATION OF BEHAVIOR

do become fearful at the sight of a hot radiator, an oncoming car,
high places, deep ditches, electric outlets, and so on.

At first, when we try to inculcate these protective fears into our
children, we cannot use language in any effective way. The words
"hot," "hurt," "burn," and "ouch" mean nothing to the infant.
During the early stages of training, therefore, most children must
actually experience the trauma and pain provided by noxious
stimuli. But with the passage of time the child experiences re-
peated pairings of warning words such as "hurt" with unpleasant
or painful sensations. These warning words can thus come to
arouse the child's anxiety or fear in the same way as do other
stimuli such as the sight of the hot stove, the sharp razor blade,
or the electric outlet. The conditioning of fear to verbal cues of
this variety is an extremely important step in providing the child
with protective fears, since by this means the parent can arouse the
child's anxiety in new situations to which the child has not been
previously exposed. The child may never have seen a soldering
iron, but the parent can elicit fear and withdrawal simply by
pointing to the iron and saying "hot" or "hurt." It would seem,
therefore, that language serves, in this manner, the very useful
adaptive function of eliciting fear and withdrawal in a large assort-
ment of potentially dangerous situations. The widespread effects
of this type of training are well illustrated by the behavior of a
)Oung child known to the writer who used the word "Ow!" as the
name for flaming matches, cigarettes, bonfires, and stoves, long
before she could give these objects their correct names. A child
need not cut his fingers on a power-driven saw to acquire a fear
and an avoidance of the whirhng blade. But to do so without
pain, he must learn to react with anxiety to certain verbal or ges-
tural symbols, and these symbols must, in turn, be associated, some-
times repeatedly, with to-be-avoided objects or situations.

If we look at the technical details of this process, it becomes ap-
parent that the initial step in associating fear with the warning
words fits the pattern of classical conditioning precisely. But in
new situations, where the child may not actually get hurt, the
paradigm is different. The sight of the new object (CS) is not
associated with pain but with the word "hurt," which, because it
can arouse fear, now functions as a substitute UCS. By repeated

LEARNED RESPONSES AS SOURCES OF DRIVE 171

pairings of the new CS with the fear-arousing word, the new CS
also comes to evoke fear and/or withdrawal. This paradigm will
be recognized as that of higher-order conditioning, a procedure
described and successfully demonstrated by Pavlov (1927) and
others. While the higher-order conditioning of salivary and skeletal
responses in animals is rather difiEcult to establish and is some-
what unstable, it may be that the higher-order conditioning of
fear can be more readily accomplished. Moreover, higher-order
emotional conditioning of a strength and stability unattainable
with animals can perhaps be produced in human subjects as a
consequence of the operation of language.

Proceeding on the premise that human behavior is fraught with
learned anxieties and fears, we must next consider whether, and
to what degree, such anxieties can lie at the root of the so-called
acquired drives for power, money, prestige, and the like.

Anxiety as an Element of the "Acquired Drive" for Money. Al-
though the paradigm of classical conditioning seems to serve
rather well for fear and anxiety, it cannot easily be applied to an
analysis of the "drives for specific goal objects" that adult human
beings are alleged to possess. Consider, for a moment, the so-
called drive for money. If this is essentially a learned or condi-
tioned response, we must first ask how it has become conditioned.
But in seeking an answer we immediately run into difficulty, since
we cannot specify any unconditioned stimulus capable of eliciting
the money-seeking drive (or response) in the first place. If there
is no stimulus that can elicit the response without training, how
can we arrange for the response to become associated with a con-
ditioned stimulus? Whatever the so-called acquired drive for
money may be, it is not a simple, unitary response elicitable by a
specific unconditioned stimulus and conditionable to other stimuli
by the procedures of classical conditioning.

Part of the difficulty here arises from the use of the phrase "ac-
quired drive for money." If, as we have suggested, it is desirable to
use the word "drive" only when referring to a construct that is
nondirective in its motivating effects upon behavior, then we can-
not meaningfully speak of a drive for any specific goal, whether
it be for money, prestige, power, or whatever. On this view, al-
though both learned and unlearned responses may contribute to

172 THE MOTIVATION OF BEHAVIOR

drive, one can have neither a learned nor an unlearned drive for
anything. While working in factories, schools, stores, offices, and
elsewhere, people do learn, of course, to perform a multitude of
responses for which monetary rewards are provided. And in a
certain sense money is the goal toward which these responses are
directed. But unless our concept of drive is changed, the drive
underlying those activities cannot itself be directed toward money;
nor does it seem likely that those responses can serve as the source
of drive to which reference is made in speaking of a "drive for
money." In fact, it is precisely these day-to-day, work-situation
activities that are themselves presumably motivated by the money-
seeking drive, and we are forced to look elsewhere, therefore, for
learned responses having the motivating properties we seek.

According to one interpretation (Brown, 1953a), anxiety might
serve as a learned motivating agency for money-seeking responses
if it is aroused by cues indicating the absence of money. On this
view, stimulus patterns such as those provided by an empty wallet
or by an "overdrawn" notice from one's bank acquire, through
learning, the capacity to elicit reactions of insecurity, uneasiness, or
anxiety. And because of the drive properties of such anxiety reac-
tions, a wide assortment of specific stimulus-elicited responses can
be affected as though some motivational variable had been manip-
ulated. Moreover, a reduction in anxiety, should it occur follow-
ing a particular response, would be expected to reinforce that re-
sponse. At the heart of this interpretation, then, is the basic idea
that an ". . . important motivating component of many of the
supposed acquired drives for specific goal objects is actually a
learned tendency to be discontented or distressed or anxious in the
absence of those goal objects" (Brown, 1953a, p. 12).

If this analysis is to carry conviction, however, it is necessary to
present the details of the process by which the reaction of anxiety
becomes conditioned to stimuli signifying "no money" or "insuf-
ficient money." We need to know why we have learned to feel
anxious, embarrassed, and uneasy if, for example, after ordering
and eating a meal in a restaurant, we find that we have failed to
bring sufficient money. By what process does anxiety become con-
ditioned to the stimuli provided by the to-be-paid check in com-
bination with the empty pocket?

LEARNED RESPONSES AS SOURCES OF DRIVE 173

Probably the most satisfactoty answer to this question is that
these anxieties are acquired through a process of higher-order con-
ditioning, with verbal cues playing a major role, in essentially the
same way that generalized fears of dangerous situations are learned.
The initial stage in the process would consist of repeated child-
hood experiences in which pain has been associated with verbal or
gestural cues of alarm or warning provided by adults. In their
most general and common form, these are statements like "Look
out, you'll get hurt," or "If you do that something bad will hap-
pen to you." Or the cues may be facial expressions of alarm, con-
cern, and worry accompanying the verbal warnings. Following the
paradigm of classical conditioning, these alarm-denoting cues
might come to arouse the emotional reactions of dread, fear, or
anxiety, even in the absence of actual pain.

The second stage of the process involves the further conditioning
of anxiety to a wide variety of cues, all of which indicate a lack
of money or its relative scarcity. Because of the nature of our
society and the great significance of money for many of its mem-
bers, ample opportunities exist therein for the occurrence of this
secondary conditioning of anxiety. This could be accomplished
in a number of ways. Consider the typical warnings given to
children by parents. "If you get on a train and can't pay your
fare, the conductor will throw you off," or, "If we run out of
money before the end of the month, we may not have enough food
to eat," or, "If you don't pay for your meal, the restaurant owner
will make you wash all the dishes." Now in each of these state-
ments, and in innumerable ones like them, an anxiety-arousing
warning is paired with a statement about the lack of money.
Through repeated pairings, by other children as well as by parents,
of these two kinds of assertions, cues denoting a lack of money
might come to arouse anxiety reactions. The cues become, in this
way, the counterparts of the fear-eliciting CS used in animal ex-
periments, and the anxieties they elicit may constitute an impor-
tant motivational component underlying human money-seeking
responses.

It is not necessary, according to this interpretation, to assume
that one must have absolutely no money in order to experience
anxiety. Being "broke" means different things to different in-

174 THE MOTIVATION OF BEHAVIOR

dividuals. The college professor may feel "broke" and anxious
when his bank balance drops below, say, $25. But the multimil-
lionaire, who owns numerous yachts, planes, estates, and cars,
may suffer acute distress upon finding that he is "down to his last
million." Just recently one of the richest men in the world was
quoted as having said that "... after all, a million dollars isn't
worth what it used to be."

One of the consequences of this view relates to the problem of
the rewarding or reinforcing value of money. As we well know,
sequences of action followed by monetary rewards tend to be
learned by individuals in our society and tend to be abandoned
if these rewards are withdrawn. One common interpretation of
this phenomenon is based on the concept of secondary reinforce-
ment, which was derived originally from Pavlov's work on higher-
order conditioning. As usually stated, a stimulus object acquires
secondary reinforcing properties when it has been closely and
repeatedly associated with a primary rewarding event. Thus if a
rat has been repeatedly fed in a white goal box, the box may act
as a learned or secondary reward in other situations because of
its association with the consummatory act of eating. Applied to
money, the principle would be that money has become secondarily
rewarding because it has been frequently exchanged for food,
which in turn has led to a reduction of hunger. The behavior of
chimpanzees in the well-known token-reward studies of Wolfe
(1936) and of Cowles (1937) has commonly been explained as
an instance of secondary reinforcement. These studies showed
that initially neutral objects, such as poker chips, acquired a
rewarding value for the chimpanzees if the chips could be ex-
changed by the animals for raisins or other preferred foods.

The foregoing interpretation of how anxiety may form the
motivational substratum of money-seeking behavior leads to a
different view of the process by which money acquires its reward-
ing value. If cues denoting the lack of money or its relative
absence lead to anxiety, then the actual possession of money must
lead to a diminution of anxiety. Moreover, if anxiety reduction
functions to reinforce antedating responses, then some of the
reinforcing value of money may be due to its anxiety-quelling
properties. Thus money may become a reinforcer, in part, because

LEARNED RESPONSES AS SOURCES OF DRIVE 175

anxiety is aroused by stimulus events denoting a lack of money
and because money in one's hand counteracts that anxiety. Inciden-
tally, this conception does not negate the value of the secondary-
reinforcement theory. It serves, rather, as an additional explanatory
mechanism, capable of operating jointly with secondary reinforce-
ment.

Money, though of major significance, is by no means the only
agent capable of eliminating the anxiety occasioned by cues
denoting money's absence. Quite often people who are distressed
by the knowledge that they are broke, or nearly so, escape from
the disquieting cues by going to sleep, getting drunk, or going
to the movies. Since these "escape" reactions are followed by
anxiety reduction, they too tend to be reinforced, and as a result
may become well established though unrealistic modes of reacting
to a lack of money.

The essence of this interpretation has been summarized by
Brown (1953ci, p. 14) as follows: "In many instances, if not all,
where adult human behavior has been strongly marked by money-
seeking responses there appears to be little need for postulating
the operation of a learned money-seeking drive. One does not
learn to have a drive for money. Instead, one learns to become
anxious in the presence of a variety of cues signifying the absence
of money. The obtaining of money automatically terminates or
drastically alters such cues, and in so doing, produces a decrease
in anxiety. Money-seeking responses, or other reactions, appearing
during the arousal of anxiety are strongly reinforced by the decline
of anxiety attending the receipt of money."

As a final point we must emphasize that this conception is not
restricted to the so-called acquired drive for money. Anxiety-arous-
ing properties could certainly be acquired by stimuli denoting a
lack of prestige or affection or power, and could thus provide a
motivational mainspring for responses directed toward these goals
as well. Moreover, this view of anxiety as a basic component of
many of the so-called acquired drives is very similar to previously
proposed conceptions. Tolman (1942), for example, has presented
a somewhat similar explanation of the "drive of gregariousness."
Thus he maintains that in gregarious species of animals, separation
from the flock or herd results in an "internal sufferance" that is

176 THE MOTIVATION OF BEHAVIOR

eliminated by rejoining the group. Clearly, this "internal suffer-
ance" is functionally much the same as the anxiety we have been
considering, and like anxiety can be aroused by cues indicating a
special kind of deficit. A parallel line of reasoning has been fol-
lowed by Dollard and Miller (1950) in their brief analysis of the
anxieties experienced by a child when separated from its mother.
Certain environmental cues, when not accompanied by the sights
or sounds of the mother, are assumed to lead to the drive of fear,
and the child's responses of seeking and approaching the mother
are reinforced by the reduction in fear her presence provides.

Other Learned Responses as Sources of Drive

Anxiety doubtless constitutes a significant secondary motivating
agency for human behavior, but there seems to be scant reason
for regarding it as the only such learned motivator. Quite possibly
other responses having motivational effects are learned by human
beings, and some perhaps by animals. According to McClelland
(1951), for example, human beings are driven as much by a
"hope of success" as they are by anxiety or by a "fear of failure."
Of the several attempts that have been made to develop theories
of acquired motivators other than anxiety, three will be considered
here. The first is the Tg-Sg "expectancy" mechanism of Hull and
Spence; the second is the "affective-arousal model" proposed by
McClelland, and the third is the writer's tentative hypothesis
concerning the motivational role played by verbal self-instructions.

The Tg-Sg Mechanism as an Acquired Source of Drive. In some
early theoretical articles, Hull (1931, 1937) made considerable
interpretive use of what is described as the "expectancy" or rg-Sg
mechanism. According to Hull, when an organism is rewarded
while making a consummatory goal reaction {^g), such as eating,
associations are formed by the process of classical conditioning
between Kg and a variety of stimuli, both internal and external.
Since some cues, e.g., those arising from the need for food (S^s),
are present throughout all phases of the instrumental response
sequence, there is a tendency for them to elicit Kq at every point
in the sequence. But only certain portions or phases of the goal

LEARNED RESPONSES AS SOURCES OF DRIVE 177

reaction of eating can be performed prior to the attainment of
food. Responses such as seizing the food, biting it, chewing, and
swallowing cannot move backward (appear at an earlier time) in
the sequence since they lack, at those points, the necessary goal
object (food) for their execution. Reactions such as sniffing, sali-
vating, smacking the lips, and swallowing, however, can indeed
be elicited before the goal is reached. Hull thought of responses
like these as split-off portions of the goal reaction (Rg) • He named
them, therefore, fractional anticipatory goal reactions and denoted
them by the symbol Vg. It should now be clear why the term ex-
pectancy is appropriate in connection with these responses. Observ-
able responses of salivating, swallowing, and licking the chops, if
exhibited by a pet dog, say, just before he is fed, constitute about
as good an objective index of his "food expectancy" as can be
found.

Now these fractional anticipatory goal reactions would be
expected to give rise to characteristic internal stimuli, and Hull
completed the picture by adding these stimuli (SgS) to the VgS.
Thus the complete rg-Sg mechanism was formed. In Hull's earliest
use of this mechanism, the Sg portion played the principal role
as a behavior determinant, serving both as a stimulus to which
a variety of reactions could become conditioned and as a secondary
reinforcer. On a number of occasions, however, Spence (195 la,
1956) has pointed out that the fractional anticipatory goal response
and its interoceptive and proprioceptive stimuli might operate
motivationally as well as associatively. Since the classically condi-
tioned rg is learned, it would qualify, by our criteria, as a secondary
source of drive, provided it does indeed have the motivational
effects suggested for it. In Spence's formulation, which we have
outlined in Chapter 4, the strength of the r^ is dependent, in part,
upon the magnitude of the food reward given an animal. And
the intervening variable K, the incentive-motivation factor, is, in
turn, a function of the strength of r^. Within the system, K affects
behavior in much the same way that D does. Thus the rg-Sg
mechanism and the related K factor are used to explain the fact
that animals will run faster through a maze if they are expecting
a large piece of food than if they are expecting a small piece. If
they have been given large pieces of food, the rg is stronger and.

178 THE MOTIVATION OF BEHAVIOR

therefore, the habit of approaching the food is multiphed by a
higher value of K.

As Spence has pointed out, a similar theory was proposed earlier
by Crespi (1944), who, in interpreting his experiments on the
effect of size of incentive, appealed to what he called an "emo-
tional drive." In this so-called "eagerness" theory, variations in
amount of reward were assumed to lead to differences in amount
of anticipatory tension or excitement, and these excitement dif-
ferences produced corresponding differences in motivation and
hence in level of performance.

There are several avenues through which anticipator}^ goal re-
sponses and their interoceptive cues might exert their motivational
effects. Within Hull's theory, for example, s^ could have a dy-
namogenic or energizing effect resembling that of intense external
stimuli. The effect of stimulus intensity, per se, Hull called
"stimulus dynamism." Or it may be, as Spence has suggested, that
conflict would ensue when the tendency for Tg to occur at an
early point in a response sequence is opposed by tendencies to
make other responses. The excitement or tension produced by the
conflict might have motivational consequences by adding an incre-
ment to the existing level of drive (D). This possibility is dis-
cussed in more detail in Chapter 6. *

An interesting application and extension of the fg-Sg concept
appears in a study by Birch, Burnstein, and Clark (1958). During
an initial 35-day training period, rats were allowed to eat dry food
pellets from 5:30 to 7:30 p.m. each day. The purpose of this
training was to develop an association between anticipatory eating
responses (r^s) and the internal cues characteristic of 22 hours of
food deprivation (Sds). On the assumption that the r^ habit
strength generalized along the dimension of So intensity, one
would expect that the r<,s evoked by the combination of external
cues and SdS would become progressively weaker with departures
of deprivation time in either direction from the 22-hour training
value. And on the further hypothesis that drive is monotonically
related to the vigor and/or frequency of TgS, drive should increase
with deprivation up to the customary time of feeding and then
decrease as deprivation is further extended. On this view, then,
the strength of the rat's food expectation (r^) is contingent upon

LEARNED RESPONSES AS SOURCES OF DRIVE 179

how similar his interoceptive cues are to those that have usually
been present at the time of eating, and his drive level is determined
by the strength of his expectancy.

Data interpreted as supporting this conception v^^ere obtained
from two behavioral measures. The first, which was the number
of times the rats depressed the empty food trough during a 46-hour
deprivation test period on the thirty-sixth and thirty-seventh days,
showed that trough depressions, though relatively infrequent during
the first 19 hours, increased rapidly during the period from 19
to 24 hours and then declined to zero at the 32-hour point, there
being some suggestion of further increases in the period from 32 to
48 hours. This is consistent with the view that the expectancy-
generated drive increased to a maximum at about the usual time
for eating and then decreased with further deprivation. Unfor-
tunately, the food troughs had been present in the cages through-
out the 35-day training period, and hence tendencies to approach
the troughs for food at deprivation times shorter than 22 hours
may have become extinguished through nonreinforcement. More-
over, the rats' relatively weak tendency to depress the troughs at
intervals longer than about 25 hours may have been due in part,
as the authors note, to extinction's taking place during the middle
periods of the 46-hour testing period.

The second behavioral measure, speed of running down a straight
alley for food, showed that on the first trial speed tended to rise
slightly, but not significantly, as deprivation time was increased
from 15 to 37 hours. On subsequent trials, however, running speeds
were markedly elevated for animals running under 22 and 25
hours of deprivation and depressed for those tested with 15 and
37 hours of deprivation. Birch et al. interpret these findings as
supporting their hypotheses concerning the conditioning of Yg to
Sd, and the contribution of r,, to D. However, all of their measures
showed that performance under 25 hours of deprivation was
superior to that at 22, despite the fact that the animals were never
fed save at the 22-hour period. If generalization of Tg along the So
dimension were the only factor involved, performance should have
been worse at 25 than at 22 hours and there should have been no
tendency for trough depressions to be greater at 46 than at 32
hours. Nevertheless, the experiment suggests that rhythmical feed-

180 THE MOTIVATION OF BEHAVIOR

ing schedules may influence privation-motivated behavior, and that
estimates of drive based solely upon time of deprivation may
need to be modified when such schedules have been employed.

Before leaving this topic, we find it appropriate to note that Mil-
ler and Bollard's (1941) analysis of acquired drives also contains
elements that are quite similar to those involved in the Tg-Sg -> K
mechanism. Thus they believe that any strong stimulus has drive
value (cf. Hull's stimulus-dynamism concept) and that the stimuli
produced by learned responses constitute the basis for all acquired
drives. The learned responses of fear and anxiety are regarded as
especially significant contributors to level of motivation (cf. Mil-
ler's experiment on fear in an earlier section of this chapter), but
responses toward such "positive" goal objects as food, money, and
sexual objects also provide drive-augmenting stimuli. The sight
of food, for example, may elicit a learned appetitive response, and
if the interoceptive stimulation attending that response is suffi-
ciently intense it will serve as a drive to impel food-seeking (and
other) responses. Moreover, all response-produced stimuli are said
by Miller and Dollard to have cue value as well as drive-enhancing
power. This aspect of their conception is identical with the notion
that a variety of responses can be associatively connected to the Sg
portion of the rg-Sg mechanism.

The Affective-arousal Model. McClelland, Atkinson, Clark, and
Lowell (1953), expanding upon some of McClelland's earlier
work (1951), have made a serious attempt to solve some of the
acquired-drive problems and have proposed what they call the
affective-arousal model as a general theor}' of motivation. According
to these writers, two affective states of the organism serve as im-
portant motivators. One of these is negative (fear or anxiety) and
the other positive (anticipation of rewards, appetite). Since both
of these motivating states are learned or conditioned reactions,
they fit the authors' definition of a motive: "A motive is the
redintegration bv a cue of a change in an affective situation"
(McClelland et al., 1953, p. 28). As shown in their example, if a
buzzer is associated with the eating of a sweet substance such as
sugar, the buzzer will eventually come to evoke (redintegrate) a
state involving positive affective change. This positive affective
condition is termed appetite. The corresponding negative affective

LEARNED RESPONSES AS SOURCES OF DRIVE 181

state, anxiety, is conditioned in the same manner as is fear, accord-
ing to our previous description. That is, a buzzer, if associated with
shock, acquires the capacity to redintegrate a negative affective
state that may have motivating effects on behavior. According to
this theory, then, an individual learns to make anticipatory affective
or emotional reactions to cues denoting the coming of both pleas-
ant and painful events. And both the positive and negative affects
are believed to have important motivating properties.

In their elaboration of this view, McClelland et al. maintain
that all motives are learned. That is, unless the tendency to respond
affectively in either a positive or negative manner is a learned one,
it does not qualify as a motive. An unlearned primary affective
state that provides the basis for affective conditioning is not in
itself a motive, but the learned or redintegrated image thereof is
a motive and is motivating. Following this line of reasoning, they
arrive at the conclusion that a newborn rat, when it experiences a
need for food (primary affect) for the first time, is not motivated,
or at least its hunger motive does not develop until it has eaten
food and the internal need-produced cues have come to function
as conditioned stimuli to redintegrate an affective arousal like that
experienced while eating. Unfortunately, McClelland and his
collaborators do not tell us whether an unlearned primary ajfect
has motivating effects even though it is not a motive. If it does,
then their assertion that the hungry baby rat does not have a
motive might be construed as having the same meaning as the
statement (Brown, 1953a) that a newborn infant, though in need
of food, cannot be said to have a drive for food. Apparently a
motive, in McClelland's terms, is roughly equivalent to Hull's
excitatory potential (E) in the sense that both conceptions are
composites of associative (habit) factors and motivating (drive)
factors. Just as Hull's E, which equals H X D, would have a zero
value if the habit strength (H) were zero, so too, McClelland's
motive would be nonexistent until a habit had been formed, even
though some primary affect (D?) were present. Identifying Mc-
Clelland's motive with E makes it easier to understand why he
insists that all motives guide behavior and are learned — this is the
H part. It is uncertain, however, whether the motivating effects
of McClelland's motives can be closely equated to D since no

182 THE MOTIVATION OF BEHAVIOR

clear descriptions are provided of u'hat z motive does in its role
as a motivational determinant of behavior.

Comparing the affective-arousal model with some of the views
already considered here is of some interest. As has been noted, the
anxiety-conditioning paradigm of McClelland et al. is the same
as that employed by Mowrer (1939), Miller and Bollard (1941),
Brown and Jacobs (1949), and others, in their treatments of the
acquired drive of fear. All of these latter writers, however, differ
from McClelland in holding that both the learned fear and the
original, unlearned pain-fear reaction have motivating power. Mc-
Clelland's positive appetitive motive resembles in marked degree
the Hull-Spence Tg-Sg mechanism. Both concepts involve learning,
both are essentially expectancies of reward, and the conditions
basic to their establishment are essentially identical. The concept
of positive affective arousal also appears to be identical with
Mowrer's (1950) emotion of "hope," which is said to be aroused
by signs of forthcoming rewards and which, like fear, can energize
response tendencies.

Verbal Stimuli as Acquired Sources of Drive. It is traditional to
assert that human behavior differs from that of animals primarily
because of man's capacity to use language. If so, then one is led
to suspect that, in so far as human motivational systems are
unique, language may be involved to a highly significant degree.
Unfortunately, httle research has been done on the relation of
language to motivation and we have few theories as to how behav-
ior may be affected by words. But since the abihty to use words
is clearly learned, it is interesting to consider the likelihood that
certain words, probably by virtue of the learned responses they
evoke, have motivating functions and that these responses, there-
fore, would qualify as acquired sources of drive.

To begin with, it is clear that some words when spoken by
certain individuals can affect the behavior of others as though a
motivational variable had been introduced. For example, verbal
commands like "Hurry-up!" "Pay attention!" "Come on now!"
and "Get ready!" when spoken by a parent to a child, tend to
facilitate a variety of different acts such as eating, reading, dressing,
walking, or doing the household chores. On some occasions these
commands can be given in a relatively soft voice and still be effec-

LEARNED RESPONSES AS SOURCES OF DRIVE 183

tive. It seems unlikely, therefore, that the motivational effects of
such stimuli can be due simply to their acoustic intensity, provided
the commands are loud enough to be heard at all. Instead, it
seems preferable to assume that these verbal commands serve as
conditioned stimuli to arouse learned responses that have mo-
tivating effects upon other responses. Viewed in this way, words
are analogous to fear-arousing CSs, and the learned reactions they
elicit are functionally comparable to conditioned fear. Just what
these reactions are is uncertain at present, but some promising
candidates for the role will be considered shortly.

A distinctive characteristic of motivating words is that they do
not, for the most part, provide cues capable of eliciting specific
goal-directed acts. In a very real sense they have no definite con-
tent. When we tell a child to hurry we often need not specify
the activities that must be performed more quickly. Such a goalless
command, therefore, may function to speed up an assortment of
very different actions. Whatever behavior is in progress, such as
running, doing arithmetic problems, bathing, or working, will tend
to be facihtated by the command "Hurry!" Because of their
widespread power to energize almost any ongoing behavior, these
motivating words bear a functional resemblance both to fear and
to the so-called primary sources of drive.

Some verbal commands, of course, like "Drink your milk!" are
quite specific and make direct reference to a definite concrete act.
Such a command may speed up milk-drinking responses, but if
this is its only effect it would scarcely qualify as a general motiva-
tor. Rather, we would tend to regard it as a specific stimulus to a
specific learned response. It fits an associative interpretation better
than a motivational one, though the intonation with which the
command is given may have motivational effects. Perhaps, there-
fore, verbal commands serve as learned motivating agents only
when they are essentially devoid of specific content.

Now if verbal commands like these, when provided by parents
and other individuals, can exert widespread facilitative effects upon
ongoing behavior, it appears reasonable to suppose that they
would also be motivating when spoken by an individual to himself.
For instance, if a person looks at his wrist watch, and sees that
only a few minutes remain before he is supposed to meet a friend.

184 THE MOTIVATION OF BEHAVIOR

he may tell himself to hurry. Presumably, these self-administered
verbal commands, because of their similarity to commands from
others, can impel the individual to walk more briskly, ride a bicycle
faster, or drive his car at a higher speed. Those who engage in
competitive activities doubtless often resort to self-exhortations
like "I must do my best," "I must give it everything I've got,"
and "I can't quit now!" Because of the nonspecific nature of these
instructions, it is possible, perhaps, for them to enhance whatever
activity is being performed by the individual. And since all in-
dividuals of a given society tend to learn the same words, the
same self-administered "pep talk" could facilitate the offensive
and defensive play of a football hero, the pole vaulting of a track
man, the cerebrations of a chess champion, or even the responses
of an expert on a television quiz program. Moreover, the assump-
tion that the commands can be self-administered means that
motivation from this source can perhaps be aroused even in situa-
tions where the individual has had no previous specific learning
experiences. Certainly people do instruct themselves in these ways
in times of stress, and the tendency to do so is unquestionably
learned. If such self-instructions can function as motivators, they
may play a major role in human behavior as secondary sources of
drive.

One of the interesting features of this proposal lies in its possible
relations to the need-achievement motive studied by Murray
(1938) and by McClelland, Atkinson, Clark, and Lowell (1953).
According to the members of the latter group, individuals differ
importantly in the degree to which they are concerned with per-
sonal success or are motivated to perform at a high level of pro-
ficiency in a variety of situations. They differ, that is, in the
strength of their need-achievement motive. An individual is said
to have an achievement motive when either positive or negative
affect results from his perception of his own performance relative
to one or another standard of excellence.

McClelland and his collaborators base their estimates of the
strength of this motive upon subjects' imaginative responses to
pictures like those of the Thematic Apperception Test. The sub-
jects are told that the test is one of creative imagination and that
their task is to write vivid and dramatic stories about each of the

LEARNED RESPONSES AS SOURCES OF DRIVE 185

pictures. The stories are scored for degree of achievement motiva-
tion by noting whether affectively toned evaluative statements are
made about the performances of individuals in the stories. McClel-
land has presented some evidence to suggest that high need-
achievers are also those who habitually strive to succeed in various
situations. If this is the case, then these should be the individuals
who have best learned to make the kinds of responses that are
consistent with their efforts to succeed. According to the view we
are exploring, these motivating responses may be, in considerable
measure, evoked by self-administered verbal commands.

Consider, for example, the student who tries to do well in every
examination or classroom situation. Let us assume that whenever
he finds himself being tested, the cues accompanying such situa-
tions elicit learned verbal responses such as, "I must try to do well,
for if I fail, my family and my sweetheart will be deeply grieved."
The phrase "I must try to do well," in this example is believed
to function as a motivator in precisely the same way as do parental
verbal commands like "Hurry!" or "Pay attention!" In short, self-
administered "try-hard" instructions such as these may constitute
the basic "acquired drive" of the high need-achiever and may be
used by him in a wide variety of circumstances.

Our case for this interpretation would be strengthened if it
could be demonstrated that high need-achievers are more likely
than low need-achievers to provide themselves with instructions
of this general sort, that all individuals are more likely to employ
these self-instructions in times of stress than at other times, and
that high need-achievers tend to use the "try-hard" instructions in
a wide variety of different situations where efficient performance
is expected of them.

Although self-administered commands have been assumed to
operate motivationally because they evoke learned responses, we
have not yet considered what these motivating responses might
be like. One possibility is that the commands or exhortations
elicit an over-all increase in bodily tension. As is well known from
the experimental studies of Bills (1927) and Courts (1942), mod-
erate degrees of muscular tension tend to facilitate not only per-
formance in motor-skills tasks but even the learning of nonsense
syllables. Thus the motivating effects of self-instructions to "try

186 THE MOTIVATION OF BEHAVIOR

hard" or "to pay attention," might be ascribed to increases in
general bodily tonus. Presumably, the steps in the learning process
through which such increases in tension come to be associated
with verbal commands could be outlined without great difficulty.

A second alternative is that the "try-hard" types of self-instruc-
tions serve as conditioned stimuli for the arousal of motivating
emotional states. Suppose a student says to himself "I must study
hard for this exam, because if I don't, I'll flunk out of school."
Clearly, these self-instructions could evoke anxiety or fear of fail-
ure, which could have motivating effects of the kinds we have
already considered. Or, if the verbal cues have become conditioned
to fractional anticipatory goal reactions, these expectancies might
provide the required motivational increments.

While the anticipation of an unpleasant event (anxiety) and
the anticipation of a pleasant event (r^ or appetite) are thus seen,
in addition to heightened tension, as possible sources of motivation
when aroused by self-instructions, in any practical situation it may
be diflBcult to tell whether one or all are functionally present.
Probably all are frequently involved in complex situations with
human beings. As McClelland and his collaborators have noted,
the anticipation of a reward for good performance is often accom-
panied by the fear that failure will negate the reward, or even
result in actual punishment.

Are Primary-need States Condltionable?

Occasionally people remark that they sometimes are not hungry
until they find themselves in a familiar restaurant, whereupon
they suddenly become hungry. Such casual descriptions cannot be
taken too seriously, but they suggest the possibility that a stimulus
might acquire the power to elicit hunger if it has been present on
numerous occasions when an organism has been hungry. If hunger
can thus be conditioned, and if it affects other responses in mo-
tivational ways, it obviously qualifies as a learned source of drive.

Although a number of early observations are consistent with
this notion of the conditionability of hunger, none is entirely
convincing. Skinner (1933), for example, believed that the phe-
nomenon had been demonstrated in his studies of running-wheel

LEARNED RESPONSES AS SOURCES OF DRIVE 187

activity in the rat. He reports that rats, during their initial expe-
riences on the wheel, when tested for six-hour periods would run
at a fairly constant rate for about the first three hours. But during
the last half of the period, their rates of running tended to decline.
If they were fed at the end of the six-hour running period for
several days, however, a "conditioned hunger cycle" was said to
have been developed, since the previously observed slowing down
now tended to disappear. Skinner reports that a number of rats
exhibited this effect, but the records of only a single animal are
presented, and apparently no suitable controls were run. The
phenomenon, therefore, though possibly genuine, need not be
interpreted as indicating that hunger has become conditioned to
cues accompanying the passage of time. Skinner's rats may simply
have been acquiring more skill in running, or cyclic feeding may
have led to the conditioning of motivating fractional anticipatory
responses which could counteract fatigue during the latter part of
the test period (cf. Birch et al., 1958).

A somewhat similar phenomenon, also suggestive of conditioned
hunger, has been reported by Slonaker (1912) and by Shirley
( 1928) in their studies of spontaneous activity in the rat. According
to these investigators, when rats are fed at the same hour every
day, they develop a tendency to become increasingly active as
the feeding time approaches, and this activity rhythm can be
altered by changing the feeding schedule. Again, however, these
observations do not constitute acceptable proof of the presence
of conditioned hunger.

Perhaps the first experiment designed specifically to demonstrate
the conditionability of hunger was performed by Anderson (1941 ) .
Since he believed that hunger could become associated with
external cues and aroused by them, he referred to conditioned
hunger as an "externalized drive." He used several groups of rats in
his studies, but only two groups are of special interest here. One of
these was given extensive training in running a multiple-unit maze
for food reward while hungry. A comparable group, that served
as a control, was given no maze training initially. Both groups were
then fed until satiated and were allowed to run through a second
maze but without food reward of any kind. The animals in the
pretrained group performed better in the second maze than did

188 THE MOTIVATION OF BEHAVIOR

the controls. From these results Anderson concluded that the
second maze, because of its general similarity to the first, served
as a stimulus to arouse a learned drive in the pretrained animals
but not in the untrained controls.

Interpretations of the Anderson experiment are somewhat haz-
ardous, since each group consisted of only five animals and no
statistical evaluations of the differences were presented. Moreover,
Siegel (1943), using a four-unit linear maze, could not confirm
Anderson's results. Siegel's animals were given satiation trials
interposed during the learning series, but neither time nor error
scores on these trials exhibited the progressive decrement one
would expect on the "externalized drive" hypothesis. As in the
case of Skinner's experiment, however, Anderson's results may
be genuine and yet due to factors other than conditioned hunger
or externalized drive. For instance, the similarity of the two mazes
could mediate the transfer of fractional anticipatory eating re-
sponses for the experimental animals, and these responses, rather
than conditioned hunger, could provide the motivation for the
second learning. Or the control animals, since they were not
handled as frequently as were the experimental rats, might have
run less well in the second maze because they were more emo-
tional. It is also clear that the experimental animals, after having
been repeatedly rewarded for running in the first maze, would
have a stronger generalized tendency to run in the second maze,
even though satiated. If so, they would tend to reach the goal box
sooner than the controls, as indeed they did, and would thus be
more immediately rewarded by the secondary reinforcing cues of
the goal box or by being removed from the maze. And finally, the
experimental animals, who had been on a deprivation schedule
for many more days than the controls, might have had a greater
bodily need for food. If so, the laboratory operation of inducing
satiation in both groups might leave a greater residual hunger drive
in the experimental rats than in the controls. Differences in resid-
ual hunger following apparent satiation — it is rather difficult,
incidentally, to satiate rats completely — might account for the
results if coupled with reinforcements provided by the goal-box
cues or by removal from the maze.

Other attempts to demonstrate the conditionability of hunger

LEARNED RESPONSES AS SOURCES OF DRIVE 189

have been reported by Myers and Miller (1954) and by Calvin,
Bicknell, and Sperling (1953). In the Myers-Miller experiment,
hungry rats were placed in the white side of the same box used
in Miller's fear-conditioning experiments and were allowed to run
into the black side for food when they touched the guillotine door.
This preliminary training was designed to associate the cues of
the white box with hunger. Three groups of rats were given 10,
30, and 70 food-rewarded trials of this kind. A fourth group of
control animals received no comparable training. To test for the
presence of an acquired hunger drive, the animals were placed,
when thoroughly satiated, in the white box, and the operating
mechanism was so arranged that if the rats pressed a small lever
the door would open, and access to the black box would be per-
mitted. Thus the problem was whether rats that had been quite
hungry in the presence of certain stimuli (white-box cues) would,
when satiated, learn to perform a new response, the only conse-
quence of which was the opportunity to escape from those stimuli.
If the white-box cues aroused a learned hunger reaction, escape
from them should have resulted in a decline in hunger that might
reinforce the learning of the new response. Since all four groups
learned the new response about equally well, the results failed to
indicate that the association of the white-box cues with hunger
had led to the formation of a learned hunger drive. Through
additional experimentation Myers and Miller were able to conclude
that the learning was perhaps motivated by a tendency to explore
unfamiliar places, but they rejected the notion that hunger, as
such, was conditioned by their procedures.

There are a number of plausible explanations for these negative
results. For one, hunger may be a slowly changing state and, there-
fore, unlike phasic responses, simply cannot be readily conditioned.
For another, the animals in running from white to black were
obviously exposed to both white and black cues when hungry, and
hence the difference between the tendencies of the two sides to
evoke hunger may have been extremely small. Or the process of
satiating the animals, prior to testing, may be an operation that
prevents conditioned hunger from occurring to any detectable
degree. If one assumes, for instance, that conditioned hunger, if it
exists, might be a response of the contractile portions of the

190 THE MOTIVATION OF BEHAVIOR

stomach, it seems likely that complete satiation would so tightly
fill the stomach as to inhibit any learned contractions.

In the experiment of Calvin, Bicknell, and Sperling, two groups
of rats were given preliminary training in a very distinctive en-
vironment, consisting principally of a triangularly shaped box with
black-and-white striped walls. The training consisted simply in
placing the animals in the box without food for 30 minutes each
day for 24 days. One group had been deprived of food for 22
hours at the time of these trials and the other had gone without
food for only 1 hour. According to the authors, the tactual and
visual cues of the box should have acquired more of a tendency
to elicit conditioned hunger in the 22-hour group than in the
1-hour group. Tests for the presence of acquired hunger were
conducted by noting the amount of food consumed by both groups
when they were permitted to eat in the same box under an inter-
mediate and identical level of food deprivation (approximately 12
hours). The four eating periods were 15 minutes long and were
spaced over a two-day period at 12-hour intervals. Under these
conditions, the formerly hungry group ate more in the triangular
box than did the other group. Unfortunately, certain control
groups necessary for an unambiguous conditioned-hunger interpre-
tation were not included in the experiment. Because of this omis-
sion it is impossible to tell whether the 22-hour group's rate of
eating was accelerated, as a consequence of its supposed acquired
drive, or whether the 1-hour group's eating rate was depressed.
Certainly, the possibility that the 1-hour animals had learned to
lie down and sleep in the box, and that this response tended to
interfere with eating, cannot be dismissed without further experi-
mentation. It should also be noted that the major result of the
experiment, however interpreted, was not confirmed by Siegel and
MacDonnel (1954) in a careful repetition of the Calvin et al.
investigation.

It would appear from this brief review of the principal literature
in this area that only one conclusion is tenable at the present
time: there is no convincing experimental evidence to support the
contention that hunger, like fear, can, after training, be elicited
as a learned response to a conditioned stimulus. Such evidence as
does exist is either inherently inconclusive or can be interpreted

LEARNED RESPONSES AS SOURCES OF DRIVE 191

with equal plausibility by other hypotheses than that of a learned
hunger drive.

Summary

We have been concerned in this chapter with motivational sys-
tems that have traditionally been described by the terms "second-
ary drives" or "learned drives." These phenomena occupy a place
of considerable importance in contemporary theories, there being
many who hold that human behavior is governed more by these
acquired drives than by biogenic needs. At the outset, however,
it was suggested that the traditional terminology be discarded in
favor of the phrase "learned sources of drive," since there appears ]
to be little reason for supposing that the effects of motivation ^k"*^ ^
on behavior are multiple rather than unitary.

From a purely descriptive standpoint, a learned source of drive
is defined by the observation that an associative variable affects
responses other than the one being learned as though a motiva-
tional variable were present. Learned responses appear to have
motivationlike effects upon other responses either because an
increment is added to a nonspecific drive, or because distinctive
cues are produced that alter habit strengths, or for both reasons. In
any event, the major tasks facing the student of acquired sources
of drive are those of discovering which kinds of learned responses
function motivationally, of delineating the laws that govern the
acquisition and extinction of these motivating responses, and of
determining the breadth of influence of motivating responses and
the mechanisms through which the effects are produced.

A major section of the chapter has been devoted to the analysis
and discussion of studies purporting to show that conditioned fear
or anxiety affects certain indicant responses as though it were a
motivational variable. Three groups of investigations of this kind
have been identified. Those in the first group show that if a refer-
ence response is evoked while a conditioned stimulus for fear is
present, the reference activity tends to be enhanced or augmented.
This is consistent with the activation criterion of motivation.
Studies in a second group have repeatedly demonstrated that fear
reduction, defined in terms of the cessation of a presumed fear-

192 THE MOTIVATION OF BEHAVIOR

arousing CS, serves as a reinforcement for the learning of new
reactions. The capacity to lead to new learning is also one of the
commonly proposed criteria for the identification of motivational
variables. A third group of experiments consists of those in which
the evocation of conditioned fear results in the inhibition or
abandonment of an indicant response. In so far as these studies
involve the procedure of arousing fear after the elicitation of the
indicant reaction, they, too, are consistent with general concep-
tions of what constitutes a motivational effect.

A summary of the variables of which fear seems to be a function
indicates that these are, in many instances, associative variables
of demonstrated significance for the conditioning of other re-
sponses. For example, number of conditioning trials, intensity of
the UCSy number of extinction trials, and distribution of practice
are all known to affect the strength of conditioned fear.

Although a large percentage of the experimental studies of fear
have been carried out with animals, there are ample reasons for
the belief that acquired anxieties also play a major role in the
motivational structure of human subjects. The direct association
of neutral and painful stimuli may account for certain human
fears, but the acquisition of most of them seems to depend upon
higher-order conditioning involving verbal cues in the roles of both
conditioned and unconditioned stimuli.

The simple classical-conditioning paradigm, though moderately
adequate for fear, cannot be applied directly to such so-called
acquired drives as those for money, affection, power, and security.
This is because one can neither identify the response that is sup-
posed to have been learned in each case, nor specify an originally
adequate unconditioned stimulus capable of eliciting the response
at the outset. It seems likely, however, that anxiety plays an im-
portant role as the motivational basis for several of these alleged
drives. In particular it is suggested that cues indicating a lack of
affection, or of power, or of money, can acquire tendencies to
elicit conditioned anxiety. Moreover, activities involved in seeking
money, power, and so on, whenever they are successful, must
inevitably counteract the anxiety-arousing cues occasioned by their
absence, and hence will be reinforced by anxiety reduction.

The final sections of the chapter are devoted to the question of

LEARNED RESPONSES AS SOURCES OF DRIVE 193

whether learned responses other than anxiety can be meaningfully
said to have motivational effects upon reference activities. Three
somewhat different answers to this question are considered. The
first, a conception developed by Hull and Spence, assumes that
consummatory goal responses become classically conditioned to
cues present at the goal, and that fractional components of the
total response complex can be elicited in advance of goal attain-
ment. In addition to providing distinctive cues, these fractional
anticipatory goal responses are assumed to add increments to the
organism's total drive level and thereby to affect other reactions
motivationally.

A second conception, McClelland's affective-arousal model, en-
tails the fundamental assumption that all motives are learned and
function both to guide and to motivate behavior. Motives are
acquired as the result of the pairing of stimulus cues with changes
in affect, the motive being a conditioned affectively toned expec-
tation of the coming of either pleasant or unpleasant events.

The third conception of learned sources of drive, proposed by
the writer, emphasizes the possibility that self-administered verbal
commands might acquire, through a process of conditioning, the
power to affect overt reactions motivationally. These verbal com-
mands may produce their effects by leading to increased general
muscular tonus, by evoking conditioned emotional responses, or
by the production of stimuli to which facilitating responses have
been attached.

In the final section of the chapter the question is raised whether
a primary need such as hunger can become conditioned to a
neutral stimulus and thereby come to qualify as a learned source
of drive. Experimental attempts to demonstrate the phenomenon
are reviewed and the conclusion reached that no convincing evi-
dence for the conditionability of hunger has yet been presented.

CHAPTER

Motivational Consequences of
Frustration and Conflict

In recent years motivational theorists have become increasingly
concerned with the possibility that the thwarting of an ongoing
response has behavioral consequences resembling those produced
by the manipulation of motivational variables. The idea that
response interference may be followed by emotional or motiva-
tional effects is not new, however, and references to the energizing
and/or disorganizing effects of thwarting abound in literary and
philosophical works. This same idea also appears in the writings
of clinical psychologists and personality theorists, where it is com-
mon to encounter the statement that failure to achieve an expected
goal may lead to an increase in emotional or other tension and
that such tension may modify subsequent behavior. In psycho-
analytic writings this general position is expressed in terms of the
thwarting of libidinal energy and its redirection into new chan-
nels, culminating in the phenomena of conversion and sublima-
tion.

We shall not attempt to trace in detail the history of the more
technical formulations of this conception, but Lewin (1931) ap-l

194

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 195

pears to have been among the first to present a relatively clear
statement of the hypothesis. Thus, in his discussion of the conse-
quences of conflict, he states that the opposed field forces in a
conflict situation, especially when there is an outer barrier, lead
to increases in total tension. Lewin did not refer specifically to
the motivational properties of this heightened tension, but some
such concept was probably intended, since augmented tension
was said to produce restless behavior which, in conjunction with
the directive characteristics (forces) of the specific situation, could
lead to affective outbursts such as fits of anger.

Some years later, Miller and Stevenson (1936) advanced a
similar hypothesis in an effort to explain some unexpected experi-
mental findings. In their investigation, hungry rats were first
trained to run down a short straight alley for food and were then
given a number of nonreinforced (frustrating) trials. During these
trials, it was observed that the animals exhibited an ". . . appar-
ent energization of certain acts such as sniffing, tossing of the head,
and cleaning of the whiskers. These acts frequently occurred in
almost explosive manner — much more vigorously than during the
first learning trials in the alley or than during the period in the
cages immediately preceding the runs" (p. 227). In discussing
these observations. Miller and Stevenson hypothesized that non-
reward may have led to a conflict between the learned responses
of eating and the responses elicited by the empty food dish. They
assumed, moreover, that the conflict-produced, proprioceptive stim-
ulation could have facilitated other actions much as hand tension
was known to facilitate the knee jerk.

Subsequent to this proposal by Miller and Stevenson, somewhat
similar hypotheses were advanced by a number of workers (e.g.,
French, 1944; Rohrer, 1949; Brown and Jacobs, 1949; Sheffield,
1950; Amsel, 1951; Brown and Farber, 1951; Spence, 1951b; Hull,
1952; Amsel and Roussel, 1952). The precise wording varies from
writer to writer, but the essential similarity of the several views
permits us to state the hypothesis in the following general form:
when stimuli normally capable of eliciting a response are present,
but the response is prevented from running its usual course, be-
havior may be affected as though a motivational variable had been
introduced. Most versions of the concept (cf., e.g., Lawson and

196 THE MOTIVATION OF BEHAVIOR

Marx, 1958) include the specific assumption that frustration adds
an increment to general drive, but as we shall indicate, the be-
havioral consequences of thwarting can often be attributed with
equal justification to the operation of associative mechanisms.

Methods of Producing Thwarting

At first sight it might seem as though the simplest of procedures
would suffice to bring about response interference and that its
occurrence could be easily and unambiguously ascertained. Unfor-
tunately, the process is not always so simple, and in any practical
experiment, therefore, the frustrating procedures must be chosen
and introduced with considerable care. Because of this necessity
it is desirable at this point to consider some of the methods that
have been used to prevent responses from running their usual
courses. Throughout the remainder of the chapter the term
thwarting will be used simply to mean response interference, while
the term frustration will be retained to denote the state, condition,
or response assumed to be produced by thwarting.

Physical Barriers as Frustrating Agents. One of the commonest
ways of producing thwarting is by the use of physical restraint of
one kind or another. With animals as subjects, these restraints
may take the form of solid barriers in a maze path, restraining
harnesses, delay chambers, treadmills, locked Skinner-box levers,
and locked food-dish covers. Such restrictive devices may partially
or completely prevent the subject's approach to a goal; they may
be introduced at different points in the response sequence; and
they may be interposed for var}'ing lengths of time. If a response
is to be thwarted and frustration is to result, blocking devices
must be introduced into situations where cues normallv capable
of eliciting the to-be-blocked response are present. And from an
ideal point of view the introduction of thwarting devices should
result in the smallest possible change in the original stimulus
complex. Thus, for example, it would be preferable to lock a
formerly unlocked food-dish cover than to prevent the animal
from having access to it by the interposition of an opaque door.

MOTIVATIONAL EFFECTS OF FRUSTRATfON AND CONFLICT 197

With the latter method many of the originally present cues are
eliminated and new ones are introduced.

Thwarting Produced by the Removal of Maintaining Stimuli. ■
A second frustration-inducing method involves the removal of
some of the stimulus objects that normally function to support
the response while leaving the remaining stimuli unaltered. For
example, food may be removed from its usual position in the goal
box of a maze or the lever may be withdrawn from the Skinner
box. Under these conditions eating and bar pressing, respectively,
are prevented from occurring, and thwarting should result since
the residual cues still tend to elicit the response. This method may
produce a more marked alteration of the stimulus complex than
the first method, but in either instance the response is physically
prevented from occurring. The two methods may be regarded as
different, because they lead either to the blocking of different
responses or to the blocking of the same response at different
points. Thus if the cover of a food dish is locked, all of the re-
sponses of lifting the cover and of seizing, chewing, and swallowing
the food are blocked. But if the cover is left unlatched and food
is removed, only the responses of grasping and ingesting the food
are thwarted.

Thwarting by the Elicitation of Incompatible Responses. A third
way of preventing the appearance of a response is to introduce
new stimuli known to be capable of eliciting reactions incompatible
with the one to be thwarted. If an individual has been trained
to lift his hand from a key when a red light comes on, the re-
sponse may be inhibited if a green light, to which he has previously
been told to press downward, is presented simultaneously with the
red. These procedures define what is typically described as a con-
flict situation. It differs from the frustration situations we have just
described in that the thwarted response is not inhibited or blocked
by external restraints but by a competitive tendency to make an
incompatible reaction. Conflict-induced thwarting of this kind
could also be achieved by first associating a given response with
one set of cues and then training the subject to perform an incom-
patible reaction to the same cues. Since this is what often happens
in discrimination-learning situations, it serves to emphasize the

198 THE MOTIVATION OF BEHAVIOR

view (Melton, 1941) that frustration is an almost inevitable ac-
companiment of all learning.

All of these response-interference methods may be used with
either animal or human subjects. But because of the language
facility of human subjects, additional techniques of thwarting
may be employed with them that cannot be applied to animals.
For example, an individual may be said to be thwarted if, after
he has completed a task given him by the experimenter, he is told
that his performance was very poor. In many such situations the
subject's goal is to obtain the experimenter's approval. To withhold
such approval or to tell the subject that he has performed poorly
is to prevent him from reaching his goal. Verbal commands may
also be used with human subjects (and after special training, with
animals) to produce a cessation of ongoing behavior, to elicit
conflicting response tendencies, and the like. Most such instances
of response blocking correspond to what has just been described
as the competing-response method, and hence no further discussion
of them is necessary.

Criteria of Thvs^arting

The above methods for preventing the initiation, continuation,
or completion of a response are commonly found in studies of
frustration, but the use of one or more of them does not guarantee
that the response has indeed been thwarted. Hence the conclusion
that blocking has actually taken place usually rests, though often
in an implicit manner, upon the satisfaction of one or more specific
criteria of thwarting. The need for such criteria can be made clear
by considering a specific example.

Suppose a dog is moving toward a bone on the other side of a
room and that when we grasp him by the collar he stops. From
such an observation we might conclude that his approach responses
have been frustrated, but it is also possible that he was going to
stop anyway at the instant we seized his collar. If the second
alternative is true, then clearly the dog's approach behavior has
not in fact been thwarted. It is improbable, of course, that the
end of a response sequence and the thwarting operation will acci-
dentally coincide in this manner on many occasions. But since

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 199

the possibility does exist, it is desirable to have definite criteria
for deciding whether a response has suffered interference.

Repetitive, Partial Responses as an Index of Thwarting. Perhaps
the best basis for concluding that a response has been thwarted
is the observation that the subject continues to make responses of
the kind that were exhibited just prior to the introduction of the
conditions designed to be thwarting. Thus if the dog in the pre-
ceding example lunges repeatedly against the restraint imposed
on its collar, his behavior serves as an index of response inter-
ference. Although the lunges are truncated or fractional parts of
the original approach behavior they provide support for the belief
that if the restraint had not been imposed, the approach would
have continued beyond the point of blocking.

Response Resumption as a Criterion of Thwarting. A second
criterion of response interference involves the observation that
the subject resumes his original behavior following the removal
of the supposedly thwarting conditions. We might, for example,
train a cat to approach a white square at the end of a short
straight alley and then teach it, in a different situation, to avoid
a black circle. If the black circle is now put into the straight alley
beside the white square, the cat may not approach to the end
as it did formerly. This change in behavior is suggestive of block-
ing due to a competing (avoidance) tendency. But we can be much
more certain that the approach response was actually blocked by
the competing response and was not momentarily weakened by
other factors, if the approach behavior reappears upon removal of
the black circle. Resumption of the original behavior thus seems
to indicate blocking. But a failure to resume the expected response
cannot be interpreted to mean the absence of thwarting. Inter-
ference might have occurred when the black circle was first intro-
duced, but if the approach tendency were weak it might have been
extinguished by even a relatively short exposure to the negative
cue. In such a case approach would not reappear following the
removal of the black circle.

Response Failure as a Sign of Thwarting. Still a third criterion
is often used in practical experiments on frustration. Though less
rigorous than the first two, it is easier to apply. Specifically, if a
subject has always, or nearly always, exhibited a concrete bit of

200 THE MOTIVATION OF BEHAVIOR

behavior in a given situation, and if we have observed this on a
relatively large number of occasions, a sudden failure to perform
in the expected manner suggests that thwarting may have occurred.
The degree of confidence we may have in reaching this conclusion
depends upon several factors. If we have observed the behavior
on each of 100 different occasions prior to the critical frustration
trial, we can be more confident that blocking has actually occurred
than if we have seen the response on only five previous occasions.
The higher the probability of response evocation, therefore, the
stronger the conviction that response failure denotes thwarting.
But it must be remembered that response failure, due to the
removal of most or all of the original conditioned stimuli, is not
a legitimate instance of thwarting. If the appropriate stimuli are
eliminated, the tendency for the response to occur will be negli-
gible, and by definition, neither thwarting nor frustration can be
produced. Thus, for example, a hungry person who has been for-
bidden to eat should be far less frustrated in a bare room than in a
food-filled restaurant. Presumably frustration will be maximal if
a response is blocked in the presence of all the original stimuli
with which it has been associated.

Associative Interpretations of Behavior
in Frustrating Situations

Although motivational theorists favor the view that response
interference leads to a state of frustration that functions, in turn,
to increase level of drive, the behavioral consequences of thwarting
can often be explained by associative rather than by motivational
principles. The kinds of behavior requiring interpretation are
usually said to include the following: trial-and-error reactions ap-
parently directed toward the circumvention of obstacles, aggres-
sive responses, escape or withdrawal responses, disorganized actions,
anticipations of thwarting, and more vigorous or more persistent
versions of the originally thwarted reactions. In broad outline,
associative interpretations of these phenomena are identical with
associative formulations of the kinds we have already evaluated in
Chapter 4. Only a few comments are needed at this point, there-

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 201

fore, to clarify the application of associative theory to thwarting-
induced behavior.

To reiterate, the conceptual core of any associative interpreta-
tion is the notion that behavioral changes are due to modified
associative strengths resulting from changes in the stimulus situa-
tion. That such a conception can be applied to behavior in frus-
trating situations becomes clear when it is realized that response
thwarting must inevitably alter external and/or internal stimuli to
some degree. Existing cues to which the thwarted response is con-
ditioned may be weakened or strengthened; new stimulus com-
ponents may be introduced, bringing with them previously learned
responses; and currently effective stimulus elements may be elim-
inated.

To take a specific example, consider Finch's (1942) observation
that chimpanzees, who customarily obtain water from a spigot
by depressing a plunger, tend to operate the plunger with aug-
mented vigor when the water supply is turned off. As Brown and
Farber (1951) have noted, several different versions of an asso-
ciative theory can be applied to the interpretation of these observa-
tions. First, the chimpanzees may have learned from previous
experiences with the same spigot or with similar ones that if
water is not forthcoming after a gentle push, it will be produced
by a strong push. On this view, vigorous responses are simply
transferred from old to new situations because of the similarity
of the stimuli in the two situations. Second, even when the ex-
ternal stimuli of the frustration situation appear to differ substan-
tially from those which the subject has previously experienced,
the transfer of learned responses to the frustrating situation could
be mediated by internal cues characteristic of the state or condi-
tion of frustration. This mechanism may provide an explanation
of the appearance of responses that seem to be maladaptive or
inappropriate to the frustration-inducing situation. Moreover, in-
ternal cues uniquely characteristic of frustration could underlie
one's ability to learn to differentiate between emotional states due
to thwarting and other states such as fear or worry. Third, within
the limitations of the frustration situation itself, behavior-modify-
ing processes are at work, that may lead to the appearance of
more vigorous movements. That is, the tendency to press the

202 THE MOTIVATION OF BEHAVIOR

plunger gently may be the dominant reaction at first. But if water
does not appear, the gentle reaction tends to become extinguished
and may be replaced by responses having initially weaker asso-
ciative strengths but greater inherent vigor.

Concerning other reactions to thwarting, such as trial-and-error
behavior, and escape or withdrawal responses, there is little doubt
but that many of these activities can also be explained without
reference to special-purpose motivational conceptions. Generally
speaking, these reactions fall within the province of the student
of learning and have usually been interpreted in the light of the
principles of transfer of training, reinforcement and extinction,
stimulus generalization, and response competition or augmenta-
tion. Still other responses, however, are often described by such
phrases as "persistent maladaptive reactions," "fixated responses,"
or by other terms carrying an "abnormal" flavor. Thus if the
behavior is identical with, or resembles strongly, activities exhibited
at an earlier stage of development or learning, it may be labeled
"regressive." But even in these instances explanatory concepts
other than those used by students of learning and problem solving
may not be needed. Some apparent instances of regression, if not
all, can be ascribed to the weakening effects of nonreinforcement
upon initially dominant responses and to the subsequent appear-
ance of weaker, previously learned, activities. Clearly, therefore,
whenever an increase in drive appears to have been produced by
blocking, serious attention should be given to the possibility that
the observed behavioral changes are nothing more than instances
of altered associative strengths. A paper by Holder, Marx, Holder,
and Collier (1957) provides numerous additional examples of the
application of associative principles to the interpretation of frus-
tration behavior.

Motivational Theories of Frustration-situation
Behavior

Purely associative conceptions may provide satisfactory accounts
of many instances of frustration behavior, but the alternative view
that thwarting leads to enhanced motivation as well as to altered
habit strengths continues to be widely supported. It is appropriate,

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT

203

therefore, before discussing experiments purporting to have demon-
strated the motivational effects of frustration, to review, for illus-
trative purposes, two rather representative motivational theories
of frustration-situation behavior.

Brown and Farber's Formulation. The principal elements and
relations comprising the Brown-Farber (1951) conception, which
incorporates many of the basic assumptions of Hull's (1943) more
general behavior theory, are summarized in the diagram of Fig. 6:1.
Beginning with the intervening variables at the center of the dia-
gram, we note that frustration (F), which is regarded as a hypo-
thetical (defined) state or condition of an organism, is assumed
to be produced when either an inhibitory tendency (I) or a com-
petitive excitatory tendency (Ec) is aroused simultaneously with
an ongoing excitatory tendency (£<,). Thus, on this view, either

Antecedent

manipulatable

conditions

Intervening
variables

Observable
response

Blocking

Nonreinforcement-

Amount of work

No. of
reinforcements >■//,

Deprivation
Noxious stimulation"

No. of
reinforcements ^jj

Fig. 6:1. This diagram summarizes the antecedent conditions and hypothetical
variables of the frustration theory of Brown and Farber (1951). The state of
frustration (F) is assumed to be produced by competition (indicated by
double-headed arrows) between an ongoing excitatory tendency (£») and
either an inhibitory tendency (J) or a competitive excitatory tendency (Ec).
These interacting tendencies are shown to depend upon their respective habit
strengths {Ho and He), drive (D), and upon the indicated antecedent condi-
tions. The consequences of frustration are shown as either an increment in
general drive (AD), or frustration-specific stimuH {Sf), or both. These factors
are assumed to have, respectively, the same kinds of motivational and associa-
tive functions assigned to them in Hull's behavior theory.

204 THE MOTIVATION OF BEHAVIOR

a negative tendency to discontinue reacting or a positive tendency
to perform an incompatible act produces frustration. The elements
at the left of the diagram indicate that both the initially dominant
ongoing tendency and the incompatible competitive one are func-
tions of their respective habit strengths {Ho and He) and of non-
specific drive (D). Drive and the habit strengths are in turn as-
sumed to depend upon the antecedent manipulatable conditions
of deprivation and noxious stimulation, and upon number of
reinforcements, respectively. Likewise, the strength of the inter-
fering inhibitory tendency is shown as varying with degree of
response blocking or interference, with nonreinforcement, and
with the amount of work involved in the execution of the re-
sponses. In this theory physical thwarting, prior to the develop-
ment of learned tendencies to avoid the barrier, leads to a state
of frustration because of the growth of inhibition which competes
with the ongoing (thwarted) tendency. Thwarting characteristic
of conflict situations is also assumed to produce frustration, since
a competitive positive tendency can presumably be as effective as
an inhibitory potential in negating an ongoing excitatory tendency.

The behavior-modifying consequences of frustration are shown
at the right-hand side of Fig. 6:1. Frustration (F) is regarded as
having two primary effects : ( 1 ) it results in an increment to gen-
eral drive (aD), and (2) it provides distinctive internal stimuli
{Sf). Within the theory, the increment to drive is assumed to
have no new functions. It serves simply as another source of drive
and thus combines multiplicatively with associative tendencies,
whether learned or unlearned, to yield enhanced excitatory poten-
tials. The frustration-generated stimuli, likewise, are assigned no
special functions other than those commonly attributed to internal
or external stimuli of any kind. Thus, as the diagram indicates,
unlearned tendencies ( Us ) to react to the frustration stimuli may
exist from birth, and/or new habit strengths can be built up to
these internal stimuli by conditioning. The final elements of the
diagram suggest that overt action will be a function of the relative
strengths of the several excitatory tendencies aroused in the specific
frustrating situation.

It should be clear from the preceding discussion that the Brown-
Farber conception is neither purely motivational nor purely asso-

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 205

ciative. It provides a place for both motivational increments and
associative changes. In this respect it resembles other, more gen-
eral, motivational theories of behavior.

Amsel's Frustration Theory. Much of this conception was
presented originally by Amsel in 1951, but the most highly devel-
oped version appeared in 1958. For Amsel, the major source of
frustration is nonreward in situations where the organism has
learned to expect a reward. Thus, when a rat enters a goal box
where it has been previously fed and finds no food in a customary
location, the animal will be thwarted, and frustration will result.
Frustration is described at one point as a "motivational condi-
tion," but it would appear from Amsel's development that frustra-
tion is essentially an implicit reaction.

In dealing with frustration as a behavior determinant, Amsel
has stressed two major points that, while superficially somewhat
different, are nonetheless related. First, the primary reaction of
frustration resulting from nonreward is assumed to lead both to an
increment in general drive and to frustration-specific internal cues.
Moreover, if the enhanced drive persists for a period of time fol-
lowing frustration, whatever actions are elicited within that period
should show augmentation, save when strongly competitive re-
sponses are evoked by frustration (or other) stimuli. As we shall
note below, Amsel and his associates have devised a number of
ingenious experiments, the results of which lend substantial sup-
port to this expectation.

Tlie second major facet of Amsel's theory consists in the as-
sumption that frustration, like other responses, can be condi-
tioned, and that an organism may, therefore, after a series of non-
rewarded trials, come to anticipate frustration. Seward (1951)
has made a similar proposal, and Holder, Marx, Holder, and
Collier (1957) ha\"e reported experimental evidence for the con-
ditionability of frustration. The specific mechanism underlying
this effect parallels Hull's rg-Sy mechanism rather closely. On the
first nonrewarded trial, the primary frustration reaction (F) is
elicited in the goal box, and through the operation of classical
conditioning it becomes associated with the stimulus complex
present in the goal box. On subsequent trials, fractional compo-
nents of the primarv frustration reaction can be elicited in the

206 THE MOTIVATION OF BEHAVIOR

alley (or elsewhere) by cues that are similar (stimulus generali-
zation) to those that were present in the goal box. These anti-
cipatory frustration reactions (r/) are thus analogous to the an-
ticipatory consummatory responses (r^s) posited by Hull. More-
over, the r/S, like the VgS, are assumed to be accompanied by their
own characteristic internal stimuli (s/s).

Although Amsel does not deny that the arousal of the Tf-Sf se-
quence may lead to an increment in drive, as Hull and Spence
have supposed to be true of the Tg-Sg sequence, the behavior-deter-
mining functions of the r/^Sf mechanism are viewed primarily as
the capacity of Sf to elicit specific inhibitory responses. For exam-
ple, in dealing with behavior in partial reinforcement situations,
Amsel assumes that frustration is inherently aversive and that
tendencies to avoid a nonrewarding goal box can become asso-
ciated with the frustration-response-produced stimuli (s/s). On
relatively early trials, therefore, performance tends to be poorer
when the response is not rewarded on every occasion, because the
S/-elicited avoidance tendencies interfere with tendencies to ap-
proach. On laier trials, however, since the animal usually does keep
on running, the approach tendencies become more and more
strongly attached to the frustration stimuli. When all reinforce-
ment is omitted during extinction, partially reinforced animals
perform more efficiently than continuously rewarded controls
because the former subjects have associated the frustration stimuli
with running, but the latter have not.

Incidentally, Amsel's assumption that frustration is both motivat-
ing and aversive is quite in accord with the views expressed here.
Specifically, to assert that an animal tends to avoid a situation
where frustration has been produced and drive has been increased
is consistent with our third criterion for identifying motivational
variables (Chapter 2). On this criterion a variable is identified as
motivational if behavior antedating its introduction tends to be
weakened or abandoned. It is not paradoxical, therefore, to say
that an organism can learn to anticipate and to avoid frustration
and yet exhibit response augmentation immediately following
thwarting.

A detailed comparison of the Amsel and Brown-Farber the-
ories would take us beyond the intended scope of this book, but it

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 207

is interesting to observe that the appearance of anticipatory frustra-
tion can be explained by the latter theory without the added as-
sumption that frustration is directly conditionable. On this view
the introduction of nonrewarded trials after an animal has been
repeatedly reinforced for approaching a goal would be expected
to lead to the development of avoidance tendencies. Through the
operation of stimulus generalization, these tendencies could be
elicited, along with approach tendencies, in the alley leading to
the goal box. Since frustration is assumed to arise from the com-
petition between one excitatory tendency and another, it follows
that frustration could thus be generated prior to reaching a goal.
Apparent instances of conditioned frustration could in this way
be explained in terms of conditioned and generalized response
tendencies which produce frustration through their competitive
interaction.

Evaluating the Motivational Effects of Thv/arting

Should one proceed on the assumption that thwarting does, in
fact, result in drive enhancement it becomes necessary to consider
the practical question of how this phenomenon can be unam-
biguously demonstrated. In principle, of course, an experiment
designed to reveal the presence of frustration-produced drive must
be essentially the same as all other investigations purporting to
show that variables are functioning motivationally. In terms of the
position taken here, this means that frustration can be said to
have led to an increment in drive when at least one of the basic
criteria for the identification of a motivational variable has been
met and when there is little reason to suppose that the observa-
tions are attributable solely to altered associative strengths.

Frustration Drive as a General Energizer. The large majority of
experiments to be considered in the following pages have been
designed on the assumption that intensified behavior should be
exhibited at or near the time of thwarting. Implicitly, therefore, if
not explicitly, most investigators have adopted a criterion for
identifying frustration or frustration effects that is identical with
our first criterion suggested in Chapter 2.

Prior to about 1950 the only evidence favoring the hypothesis of

208 THE MOTIVATION OF BEHAVIOR

frustration-induced drive ^^'as casual and nonsystematic. Early
workers in the field of learning frequently noted that when moti-
vated subjects encountered obstructions in problem-solving and
trial-and-error situations, highly energetic activity was the result.
Thorndike (1898), for example, reported that cats in attempting
to escape from the confinement of a problem box struggled with
"extraordinary vigor." Likewise, Hamilton (1916), during his
extensive comparative studies of reactions to insoluble problems,
often observed that his subjects exhibited excited, emotional be-
havior. Moreover, an increase in response strength during the
first few extinction trials has been reported for classically condi-
tioned responses by Switzer, 1930, Hilgard and Marquis, 1935,
and Hovland, 1936. Whether this phenomenon constitutes a
legitimate instance of frustration drive is uncertain, although
omitting the UCS may strengthen competitive inhibitory tend-
encies and thereby induce some measure of frustration.

Observations of increased response vigor following the thwarting
of a well-learned response appear in increasing numbers during
the third and fourth decades of the present century. For example.
Skinner (1932) found the rat's rate of eating to be higher after a
temporary period of enforced waiting than immediately prior to
the beginning of the frustration period. Skinner used the term
"emotional" in describing this effect, but he seems not to have
considered the heightened rate to be a frustration-drive phe-
nomenon. Miller and Stevenson (1936) and Finch (1942), whose
studies we have already mentioned, reported a marked intensifica-
tion of certain responses following nonreward in a situation where
reward had formerly been consistently obtained. Others, such as
Crespi (1944), had also observed intensified activity following
both nonreward and a shift from large to small reward, and Brown
and Gentry (1948) reported numerous instances in which periods
of delay wer^ preceded by heightened emotional behavior.

Perhaps the earliest experimental study directed specifically
toward demonstrating the presence of frustration drive and toward
the discovery of the variables on which it depends is that by
Marzocco (1951). Basing his theorizing, in part, upon the sugges-
tions of Brown and Farber (1951), Marzocco reasoned that the
hypothetical state resulting from thwarting would function as an

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 209

irrelevant drive to raise the organism's total effective drive level.
Moreover, this heightened drive would be expected to increase
the amplitude of any response evoked during thwarting, provided
stimuli accompanying frustration did not lead to excessively
strong competing responses. He also hypothesized that frustration
would increase with degree of hunger and with amount of practice
in making the frustrated response. And since the frustration state
might be relatively transient, increasing the length of the interval
between successive frustrated responses should weaken the effect
of a previous frustration upon a subsequent reaction.

To test these predictions, Marzocco trained 81 rats in a modified
Skinner box in which measures could be obtained of the force with
which the rats depressed the bar. The general procedure involved
a number of days devoted to habituation and taming followed by
one day of training in bar-pressing and by a second day of testing
for frustration-drive effects. On the frustration day, all animals
were given three rewarded trials followed by a series of con-
sistently nonrewarded (frustrating) trials. Three subjects were
assigned at random to each of the cells of a 3 X 3 X 3 factorial
design. The variables of the design and their levels were: (1 ) hours
without food at the time of frustration: 1, 16, and 22; (2) number
of rewarded bar-pressing trials on the acquisition day: 8, 24, and
72; and (3) time (in seconds) between successive nonrewarded
trials on the frustration day: 10, 20, and 40. Marzocco's measure of
frustration-drive effects was a difference score obtained for each
rat by subtracting the mean force it exerted on the bar during the
first four trials of the frustration day from the mean force exerted
on the next four. When the responses for all 81 rats were grouped
together, the rise in force due to thwarting proved to be very
striking and highly significant. Figure 6:2 shows this effect quite
clearly. Here it will be seen that immediately following the fourth
trial (the first nonrewarded one) the average force rises from 28
to 35.9 grams. Moreover, if 28 grams is taken as a reference level,
it is apparent that in spite of continued nonreward the rats con-
tinued to press the lever with "supernormal" vigor until about the
seventeenth trial.

When the data for the differently treated groups were analyzed,
Marzocco found the frustration effect to be significantly related to

210

THE MOTIVATION OF BEHAVIOR

0 4 8 12 16 20 24 28 32 36 40 44
Trials

Fig. 6:2. Mean force in grams exerted on a lever by rats during a series of
trials in which the first three trials were followed by food reward and the re-
mainder were not. The marked increase in vigor of response after the first
nonrewarded trial (trial 4) provides a clear example of the energizing effects
of thwarting upon subsequent responses. {From Marzocco, 1951.)

hours of food deprivation at the time of frustration, with the
frustration-produced increase in bar-pressing force rising with de-
gree of hunger. He also found an increasing frustration effect as
the intertrial interval was shortened, though the trend was not
statistically significant. This relation had been predicted on the
ground that the frustration generated on one nonrewarded trial
should have a greater effect on the vigor of the following response
when the interval between them was short. The frustration effect
was also expected to increase directly with number of rein-
forced trials. Both the 24- and 72-reinforcement groups showed
a greater increase in force than did the 8-reinforcement group,
but the 72-trial group showed less of the effect than did the 24-trial
animals.

Taken as a whole, Marzocco's study provides clear evidence for
increased response vigor following thwarting. But his study does
not prove that thwarting leads to an increment in drive. During
their preexperimental history his animals could have learned in a
variety of situations to exert more vigorous responses in order to
circumvent barriers or to overcome resistance to the completion
of a response. If so, and if the bar-pressing situation could be re-
garded as sufficiently similar to the situations in which such re-

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 211

sponses had been learned, the transfer of the learned reactions
to the new situation could be explained. Alternatively, the animals
might have learned during their bar-pressing acquisition trials that
if no reward followed a weak response, a more vigorous press
would successfully activate the food-release mechanism. Conse-
quently, when no food appeared during extinction trials, its ab-
sence would provide cues tending to evoke more vigorous responses.
If this were the correct interpretation, however, the experimental
variable of number of reinforcements should have been more
clearly related to the frustration effect. Finally, one might even
look upon Marzocco's results as an instance of regression. His data
show that the mean force exerted on the bar by a group of 27 rats
during acquisition was 33.5 grams for the first 10 rewarded trials
and 20.6 grams for the last 10 (of 72) trials. Thus the animals were
apparently learning, during their bar-pressing trials, to exert less
and less force on the bar. If pressing the bar vigorously is regarded
as the "earher habit" and weaker pressing as the "later habit,"
then the animals' frustration behavior might reasonably be de-
scribed as "regression" to an earlier form of activity. These various
interpretations of Marzocco's results reemphasize the point that
experiments purporting to demonstrate the frustration-drive effect
must be done with great care if the credibility of associative con-
ceptions is to be minimized. That this is rather difficult to accom-
plish will become clearer as we examine additional studies of the
frustration-drive phenomenon.

Amsel and Roussel's (1952) experiment constitutes an ingenious
attempt to demonstrate the energizing effects of nonreward upon
locomotor behavior in the rat. Initially, these investigators trained
their rats to run down a short straight alley into a goal box for
food and then out into a second connecting alley and to a second
goal box for a second bit of food. Preliminary training consisted of
84 trials, on each of which the rats obtained food in both goal
boxes. During an additional 36 trials, food was omitted from the
first goal box on a randomly selected half of the runs. These 18
nonrewarded trials constituted the frustrating trials. The effects of
thwarting were evaluated by measuring the rats' speed of running
in the second section and comparing the values obtained on frus-
trating (nonrewarded) trials with those obtained on rewarded trials.

212

THE MOTIVATION OF BEHAVIOR

On all trials food was present in the second goal box. Tlie results
of these manipulations upon the median running times of the 18
subjects in the second alley are summarized in Fig. 6:3. Here it
will be seen that the rats' running time, when reward was provided
in both goal boxes, had reached a stable level at the end of the 28
days (3 trials daily) of preliminary training. Shortly after the in-
troduction of the frustration trials, however, running time on those
trials dropped to a new level, which proved to be significantly lower
than the level maintained on further reward trials. Amsel and
Roussel concluded from these data that frustration due to non-
reward in the first goal box produced an addition to the motiva-
tional complex, which resulted in the establishment of a new
and higher maximum speed of running.

Taken at face value, this finding supports the frustration-drive
hypothesis, but as Amsel and Roussel point out, other interpreta-
tions are possible. For example, they note that the difference in
the performance of the animals on the rewarded and frustration

.i 100

■319

Reward

I I I I I L.

I ' I u

I I r I [

8 12 16 20

Preliminary reward

days

24 28 123456789
Reward-frustration

test trials
Two-trial blocks

Fig. 6:3. Data indicating that rats tend to run faster on nonrewarded (frustrat-
ing) trials than on rewarded trials when the two kinds of trials are admin-
istered alternately following a long period of training during which every trial
has been rewarded. [Adapted from Amsel and Roussel, J 952.)

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 213

test trials could mean either a depression of performance on re-
warded trials or heightened performance on nonreward trials.
Speed of running during the preliminary reward days and on the
rewarded test trials might have been reduced because the animals
had eaten food in the first goal box just before entering the sec-
ond alley. If this were correct, the frustration-drive hypothesis
would not be needed to explain faster running on nonrewarded
trials.

Amsel and Roussel suggest, however, that two factors support
the argument against this performance-depression hypothesis. First,
the omission of customary reward in the first goal box should have
altered the stimulus complex (taste of food in the mouth, for ex-
ample) to which running had been conditioned, and this would
function to reduce running speed on nonrewarded trials. Second,
experimental evidence from other studies shows that a small
amount of prefeeding may increase the level of performance. Since
both of these factors would have an effect opposite to that de-
scribed by the performance-depression explanation, Amsel and
Roussel do not consider that interpretation the only reasonable
one.

Support for the wisdom of their view has recently been obtained
by Wagner (1959) . In a similar experiment, where the frustration-
drive effect was clearly demonstrated, he ran an additional control
group, whose members were detained for a few seconds in the first
goal box but were never fed therein. On the response-depression
hypothesis, these animals, whose hunger had not been reduced
just prior to running in the second alley, should have run faster
in that alley than rats that had been fed in the first goal box.
However, these control subjects of Wagner's ran at about the
same speed in the second alley as did the animals that consumed
a pellet in the first goal box.

Before turning to other types of studies purporting to demon-
strate the frustration-drive effect, we should note that Amsel and
his associates (e.g., Roussel, 1952) have confirmed the main find-
ing of the Amsel and Roussel study on a number of occasions.
Moreover, they have also shown (Amsel and Hancock, 1957) that
the frustration effect is more marked if the first alley is similar to
the first goal box than if it is different. This observation is consist-

214 THE MOTIVATION OF BEHAVIOR

ent with their assumption that frustration is a consequence of the
thwarting of an expectancy (rg-Sg) for a goal object. If the first
alley is the same color as the first goal box it should elicit a stronger
anticipation of food than if the two are different, and the frustra-
tion of the stronger expectancy should lead to a bigger increment
in drive with a greater enhancement of running speed in the sec-
ond alley.

Experimental attempts to demonstrate the presence of frustra-
tion-drive effects on human subjects, particularly children, have
been reported quite frequently in recent times. It is impossible
here to review each of these studies in detail, but those by Haner
and Brown (1955) and by Holton (1956) serve as excellent repre-
sentative examples.

One of the purposes of the Haner-Brown study was to try to
extend and clarify the frustration-aggression hypothesis of Dollard,
Doob, Miller, Mowrer, and Sears (1939). According to these lat-
ter writers, aggression is a function of, among other things, degree
of instigation to action. But Haner and Brown regard instigation
to action as being equivalent to Hull's excitatory potential, which
includes both habit and drive, and hold that Dollard et al. fail,
in their handling of this concept, to recognize the importance of
the habit strength component. They make the prediction, there-
fore, that if behavior is frustrated near the goal, where habit
strength is presumed to be stronger, the "amount of disturbance
experienced" will be greater than if the sequence is interrupted at
some point farther from the goal. Moreover, intensity of aggressive
action should vary directly with "amount of disturbance experi-
enced." In the terminology of this chapter, Haner and Brown are
asserting that degree of frustration is directly related to habit
strength at the time of thwarting and that the motivational incre-
ment is, in turn, a direct function of degree of frustration. This,
it will be recalled, was essentially the position taken by Marzocco.

In attempting to test this assertion, Haner and Brown gave ele-
mentary school children a serial task in which they had to place
36 marbles, one at a time, into 36 holes in a board. They were
told that four successful completions of the task would bring a
prize but that the time allotted to each trial would be limited and
variable. On certain trials the experimenter operated a mechanism

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT

215

which caused whatever marbles had been placed on the board to
drop though the holes into a box below. On these frustration trials
a buzzer would sound and the subjects had to depress a spring-
loaded plunger to turn off the buzzer and initiate a new trial. De-
gree of frustration-induced drive was inferred from the amount of
downward movement of the plunger. On various trials all subjects
were frustrated at four points of proximity to the goal, that is,
after 9, 18, 27, or 32 marbles had been placed in the holes on the
board. The buzzer was also sounded after all 36 marbles had been
used. But this, by definition, was not a frustration trial, since the
goal (or a subgoal, at least) had been reached.

The results of the plunger-pushing measurements obtained by
Haner and Brown are shown in Fig. 6:4. As this figure shows,
when the point of frustration was moved closer and closer to the
goal, as expressed by the number of marbles placed on the board,
the mean amount of plunger depression increased progressively,
reaching a peak at the 32-marble point, which was eight-ninths of
the distance to the goal. But the displacement of the plunger
dropped markedly when all 36 marbles had been placed and the
goal had been reached. According to the authors, the increase
in plunger force from the 9-marble point of interruption to the
32-marble point was highly significant statistically. Consequently,
their hypothesis concerning the increase in motivational effects of
frustration with increasing habit strength tended to be supported.

One of the noteworthy features of the Haner-Brown study is

Fig. 6:4. Mean amplitude
of plunger-depressing re-
sponses made by children
following the induction of
frustration at varying dis-
tances from a goal. Since
the task was to place 36
marbles in holes in a
board, the distance from
the goal decreased with in-
creasing numbers of mar-
bles placed. {Adapted
from Haner and Brown,
J955.)

18 27 32

Marbles placed in board

216 THE MOTIVATION OF BEHAVIOR

that the response of plunger pressing, from which motivational
eflfects were inferred, was clearly different from the thwarted instru-
mental response of marble placing. Because of this, an interpreta-
tion involving transfer of training rather than increased motiva-
tion does not seem highly plausible. On such a view, one would
have to suppose that the children had learned in very similar situa-
tions that vigorous plunger-pressing responses or closely similar
ones would enable them to circumvent or remove a thwarting
agent more quickly than weak plunger presses. Some children, of
course, might have had an opportunity to acquire such habits.
But generally speaking, when an explanation rests upon the prin-
ciples of stimulus generalization, regression, and the like, its plau-
sibility is reduced if the indicator response differs from the
thwarted instrumental response and if the indicator response is
causally unrelated to the removal or circumvention of the barrier.

Holton (1956) has also obtained evidence favoring the hy-
pothesis that frustration-drive effects become greater with in-
creased nearness to a goal and with increasing numbers of rein-
forcements of the thwarted response. In the initial phase of her
experiment, three groups of preschool children were taught to
make a simple spatial discrimination. Two identical orange-colored
stimulus patches were presented against a black background, and
the child's task was to push against one of the patches. A correct
response was followed by a marble reward, and a prize could be
won by accumulating a stated number of marbles. Thwarting was
induced by failure to reinforce the panel-pressing response with
a marble. The apparatus was constructed so that accurate measure-
ments could be made of the force with which each of the stimulus
panels was pushed by the subjects during both rewarded and
frustration trials.

Three groups of subjects were used by Holton. The members of
one were permitted to make only 1 3 correct responses before being
thwarted near the goal. A second group was also frustrated near
the goal, but after making 26 correct responses. The third group
was given 26 reinforced trials, but nonreward was introduced at
a greater distance from the goal than in the case of the first two
groups.

To evaluate the frustration-drive effects, Holton compared the

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT

217

mean force exerted on the four trials preceding nonreward with
the mean force on the four trials following nonreward. The values
obtained from these measurements for the three groups are
plotted in Fig. 6:5.

As this figure shows, all three groups pushed harder on the
stimulus panels on the first four nonrewarded trials than they did
on the preceding rewarded trials. Actually, only one of the 45 sub-
jects in the three groups failed to exhibit a higher mean score for
the nonrewarded trials than for the rewarded ones. Thus the
finding that nonreward leads to more vigorous responses receives
further confirmation. Holton also found, as the figure indicates,
that of the two groups blocked near the goal, the one given 26
reinforcements prior to blocking exhibited a significantly greater
increase in response amplitude than the 13-trial group. In addition,
the group blocked near the goal ( 26-N ) showed a greater increase
in force than the comparably trained group (26-F) that was
blocked farther from the goal. This is consistent with the results

Pre-thwarting
tests

Post-thwarting
tests

Fig. 6:5. Mean force exerted on response panels by children before and after
thwarting. The curve marked 26-F denotes the responses of a group given 26
reinforcements and blocked far from the goal. The curves labeled 26-N and
13-N apply to groups blocked near the goal after 26 and 13 reinforcements,
respectively. {Adapted from Holton, J 956.)

218 THE MOTIVATION OF BEHAVIOR

of the Haner-Brown experiment and with theoretical expectations.

Reduction in Frustration Drive as Reinforcement. The investi-
gations reviewed in the preceding section all point to the conclu-
sion that response enhancement at or near the time of thwarting
fits the view of thwarting as a source of drive. It is also reasonable
to suppose, if we bear in mind our second criterion for the identifi-
cation of a motivational variable (Chapter 2), that a reduction in
frustration drive should be reinforcing. Thus one would expect
that the strength of a response would increase if that response were
followed by a reduction in frustration.

As is sometimes the case with ideas such as this, the supporting
evidence, though plentiful, is almost entirely anecdotal. In the
typical trial-and-error situation, for instance, an organism tends to
learn whatever response is successful in leading to the circumven-
tion of the thwarting circumstances or to the correct solution of
the problem. Such learning may be due, in part, to a reduction in
frustration-produced drive following the performance of the cor-
rect reaction. But in these situations the problem-solving response
nearly always leads to the receipt of food or to other reinforcers.
This makes it difficult to decide whether the escape reaction is
being reinforced by a reduction in frustration drive or by some
quite different agent. There are instances, however, such as that
reported by Guthrie and Horton (1946), in which animals learn
to escape from a problem box by performing a specific act, yet do
not consume the reward when they get out. In such cases the rein-
forcement may be provided by a decrease in frustration drive since
hunger remains unaffected.

^^Negative Effects of Increases in Frustration Drive. Our third
criterion for the identification of a motivational variable leads to
the expectation that reactions followed by increased drive due to
thwarting should tend to be weakened. As noted above, Amsel's
supposition that frustration is an aversive condition is consistent
with this view.

Probably the clearest evidence for the negative effects of thwart-
ing upon prefrustration behavior is provided by an experiment of
Holder, Marx, Holder, and Collier (1957). These investigators
trained hungry rats to run down a 90-inch straight alley for food
reward. During a series of 25 such training trials, administered one

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 219

per day, the rats were detained for 1 second in a delay chamber at
the middle of the runway. On subsequent test trials, one-third of
the subjects continued to run under the same conditions as during
training, another third were detained for 15 seconds in the delay
box, and the remaining third were delayed for 45 seconds. Both
running- and starting-time measures revealed that the increased
delay led to better performance in the post-delay-box section of
the alley — this supports the concept of frustration drive as an
energizer — but to poorer performance in the pre-delay-box seg-
ment. Granting the plausibility of our third motivation-variable
criterion, we may conclude that this latter effect, like the former,
supports the assumption that thwarting serves as a source of drive.
As Holder et al. point out, the reduction in response strength
in the pre-delay segment can be explained, within Amsel's theory,
as due to the growth of an anticipatory frustration reaction or,
following Brown and Farber, as a consequence of the conditioning
of avoidance responses to the noxious cues accompanying frustra-
tion. Further experimental evidence, indicating that delays lead to
the weakening of pre-delay responses can be found in the studies
of Cooper (1938), Gilhousen (1938), Brown and Gentry (1948),
and Holder (1951).

Conflict as a Source of Drive. Throughout the foregoing pages
we have proceeded on the assumption that frustrating conditions
such as physical blocks, delays, and removal of rewards may all
involve the development of tendencies that are antagonistic to the
tendency being thwarted. And since the simultaneous arousal of
competitive tendencies is also the definitive feature of conflict situa-
tions, no attempt has been made to distinguish between frustra-
tion and conflict. In so far as conflict and frustration cannot be
differentiated, experiments purporting to have demonstrated that
conflict leads to an increment in drive would necessarily be of the
same general sort as those designed to detect frustration-generated
drive. Studies by Hollenberg and Sperry (1951), and by Lowell
(1952cz), which rest on unpublished theoretical developments of
J. W. M. Whiting, serve as examples of such research.

In concluding this chapter, we find it interesting to note that
the conflict-drive hypothesis has been extended from individual
behavior situations to social and cultural phenomena by Whiting

220 THE MOTIVATION OF BEHAVIOR

and Child (1953). These authors tabulated the child-rearing prac-
tices of 75 primitive societies and then obtained ratings of "initial
satisfaction" and "socialization anxiety." The term initial satisfac-
tion refers to the positive (pleasant or rewarding) consequences
of the socialization process at the individual level. Socialization
anxiety, however, is the consequence of severity of discipline during
child training and is approximately equivalent to an acquired-fear
reaction. Ratings on each of these two variables were obtained for
each culture in a number of different areas of parent-child interac-
tion such as aggression, weaning, and toilet training. The ratings
of initial satisfaction and socialization anxiety were then used by
Whiting and Child to test a number of hypotheses about relations
between child-rearing practices and various aspects of adult per-
sonality patterns.

The concept of conflict-produced drive is introduced in Whiting
and Child's treatment of the origins of the fear of others. Here it
is suggested that fears of human and animal spirits derive in con-
siderable measure from anxieties about the expression of aggres-
sion. Individuals acquire anxieties concerning aggressive acts to-
ward their parents during childhood as a result of the parents'
punitive role in the socialization process. These fears of parents
and near relatives, primarily, are assumed to generalize to other
people and/or spirits (human or animal) to the degree that the
generalized stimulus patterns resemble the persons originally re-
sponsible for frustration and punishment. From hypotheses such as
these Whiting and Child predict, and confirm their prediction
from child-rearing and other data, that members of primitive socie-
ties should exhibit a greater fear of human spirits than of animal
spirits. Moreover, by the use of the conflict-drive hypothesis, they
successfully predict that societies with high aggression anxiety will
have more nearly equal tendencies to fear human and animal
spirits than will societies with low aggression anxiety. The precise
steps in this and in other deductions made by Whiting and Child
are too detailed for presentation here; but it is significant that
their conflict-drive concept leads to reasonably clear-cut, testable
predictions, which apparently could not have been made without
its use.

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 221

Summary

The problem of whether an increase in general drive is produced
by frustrating or conflictful situations serves as the focal point for
the discussions of this chapter. When recast into a form con-
sistent with the conceptual framework of earlier chapters, this is
the problem of whether a reference reaction, if it is elicited at
about the time some other activity is thwarted, will be affected as
though a motivational variable had been introduced. Throughout
the chapter the terms thwarting and frustration are used, respec-
tively, to designate the fact of response interference and the state
or condition produced by interference.

A variety of methods may be employed to prevent the initiation,
continuation, or completion of a response. In frustration experi-
ments with animal subjects, physically restrictive devices are com-
monly used. These may take the form of solid barriers in a maze
pathway, delay chambers, restraining harnesses, and locked food-
dish covers. Thwarting may also be produced by removing some
of the stimulus objects upon which the to-be-thwarted response
depends or by introducing new stimuli to which incompatible reac-
tions have been, or may become, conditioned. With human sub-
jects, thwarting can be produced by verbal instructions to the effect
that the subject has failed or has performed at an unsatisfactory
level. In general, the ideal method for producing thwarting would
seem to be one that interferes with a response or response chain
while producing the least possible alteration of the normal, re-
sponse-eliciting stimulus complex.

The introduction of experimental conditions believed to be
thwarting is not alone sufficient to insure that the to-be-blocked
response will indeed suffer interference. Typically, therefore, the
success of a blocking operation is inferred from one or more overt
behavioral changes. The observation that a subject repeatedly
makes partial responses resembling his prethwarted reactions
serves as one criterion for evaluating the success of thwarting
operations. Abrupt response failure following the introduction of a
frustrating condition and response resumption upon the removal
of that condition may also indicate successful response blocking.

222 THE MOTIVATION OF BEHAVIOR

Motivational theorists, for the most part, are partial to the view
that thwarting induces a state (sometimes a response) of frustra-
tion, which in turn raises the subject's level of drive. Nevertheless,
interpretations containing no reference to drive as such may
also provide satisfactory accounts of certain phenomena charac-
teristic of thwarting. In general, such explanations of frustration-
situation behavior rely on the view that thwarting must necessarily
modify stimulus conditions and hence associative strengths. In-
creased vigor of responding, persistent trial-and-error behavior, and
other activities typical of frustrating situations are thus seen as
instances of transfer of training, stimulus generalization, and the
learning of new responses to cues attending thwarting.

Writers favoring motivational views generally concur in as-
suming that frustration leads both to an increment in drive and
to the production of frustration-specific stimuli. For some authors,
however, frustration assumes the systematic status of a response.
This leads to the conclusion that frustration can be conditioned
and can therefore be elicited in complete or in fragmentary form
by appropriate stimuli. The weakening of instrumental reactions
antedating frustration is explained on this view as an instance of
the occurrence of fractional anticipatory frustration reactions hav-
ing inhibitory properties.

In the majority of experiments designed to reveal the motivat-
ing properties of frustration, an increase in the strength of an
indicant response has been taken as the criterion of heightened
drive. Thus, enhanced running speed following nonreward or delay,
heightened vigor of lever pressing following extinction trials, and
excited or exaggerated movements have all provided support for
the frustration-drive conception. The weakening or abandonment
of responses antedating an increase in drive is also a potentially
useful criterion for the identification of motivational variables.
Experimental data from studies in which delays are introduced
in runways or mazes are coordinate with this criterion, since delay
typically leads to poorer performance in pre-delay portions of a
maze. Systematic data supporting the supposition that a reduction
in frustration drive functions as reinforcement for the learning of
new responses are conspicuously absent.

Thwarting may usually, if not always, give rise to the appearance

MOTIVATIONAL EFFECTS OF FRUSTRATION AND CONFLICT 223

of competitive reaction tendencies, and since conflict is commonly
defined by reference to response competition, no attempt has been
made here to differentiate between frustration and conflict. Some
experimental findings suggest that situations labeled "conflictful,"
like those termed "frustrating," may generate increases in level
of motivation.

CHAPTER

Motivational Variables and
Hunnan Perfornnance

In the preceding chapters the major portion of the experimental
and illustrative material was drawn from investigations of animal
behavior. This procedure can be justified on the ground that in
large measure the problems and theoretical issues under examina-
tion arose from basic experiments with animal subjects and that
these investigations have been more extensive and precise than
studies of human motivation. Significant contributions have been
made, of course, to our knowledge of motivation by research
workers who have been concerned exclusively with human be-
havior. But human beings, probably because of the extensive role
that language plays in their activities, present us with unique
motivational problems requiring special consideration. This and
the following chapter, therefore, are exclusively devoted to the
analysis of these problems. Although a sharp distinction probably
cannot be drawn between behavior that is perceptual and behavior
that is not, material bearing on the relations of motivational vari-
ables to performance in perceptual tasks has been treated sep-
arately in Chapter 8.

224

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 225

In its organizational structure, this chapter parallels our earlier
discussion of methods of defining drive. Thus, in Chapter 2, the
assertion was made that drive can be independently defined by
reference to antecedent conditions, to stimulus conditions, to or-
ganic states, or to subjects' responses in standardized test situa-
tions. Here, therefore, we shall consider some of the effects on hu-
man performance of drive-level variations believed to result from
deprivation and from the administration of strong stimuli, as well
as the effects of drive differences defined in terms of individual
differences in test-situation behavior.

Deprivation-induced Motivation

Because of practical difficulties attending the induction of
severe deprivation states in human subjects, research studies of the
effects of such deprivations on performance are rare. Moreover, of
relevant experiments, a majority have involved behavior that is
typically termed "perceptual" and hence are discussed in the fol-
lowing chapter.

On purely theoretical grounds, whether of the motivational or
associative variety, one would anticipate that specific behavioral
changes would be exhibited when human subjects are deprived of
food. For example, from the multiplicative-drive theory it would
be predicted that food deprivation would lead to an increase in D
and hence to an increase in the strengths of all excitatory tend-
encies. Performance level in a variety of situations should be im-
proved by hunger, therefore, provided only that the tendencies to
perform the correct responses are the dominant ones in the hier-
archy of reactive tendencies. However, increased drive should lead
to poorer performance whenever incorrect tendencies are stronger
than correct ones. Coupled with the effects of increases in D are
effects attributable to changes in internal stimuli {Smvs), includ-
ing self-administered verbal commands or instructions. With ex-
tended periods of deprivation, these cues may increase in intensity,
numerosity, or persistence, and since specific correct or incorrect
response tendencies may be associated with these stimuli, marked
changes in the relative strengths of all reactive tendencies could
accompany severe deprivation. As a consequence, definite predic-

226 THE MOTIVATION OF BEHAVIOR

tions can be made only when knowledge of the comparative
strengths of task-relevant and task-irrelevant tendencies is avail-
able. Behavior is jointly determined by external and internal
stimuli, however, and hence the effects of deprivation should also
depend, to a considerable degree, on the kinds of stimuli provided
by the behavior-testing situation and on their relations to the
subjects' learning history. Thus a stimulus such as a picture of
spaghetti would presumably be more likely to elicit food-related
responses from a hungry Italian child than from a child who has
had fewer opportunities to associate such responses with both
internal hunger stimuli and the visual cues of spaghetti.

The view that hunger should, as a consequence of intensified
internal cues, lead to an increase in food-related verbal responses is
supported in some degree by both anecdotal and experimental evi-
dence. Historical reports of personal experiences during periods of
famine or extended deprivation (Sorokin, 1942) indicate that at
such times concern with food becomes a topic of overwhelming
importance, toward which the individual's thoughts and activities
are entirely and unceasingly directed. For the person who is starv-
ing food becomes the central focus of daydreams and conversa-
tions.

Of especial interest in this connection is a report by Brozek,
Guetzkow, and Baldwin (1951) of a study of semistarvation con-
ducted during the year 1944-1945. The subjects of this experi-
ment, who were conscientious objectors, volunteered to undergo
a period of greatly reduced food intake for a period of 24 weeks.
Consistent with the reports of ill-fated expeditions by explorers
and others, the 36 semistarved subjects gave clear evidence that
much of their waking thoughts centered on food and on related
matters such as cooking and agriculture. "They talked and read
about food. Their attention was attracted by the scenes of food
and eating in the movies. . . . Almost any discussion was likely
to end by talking about food. Cook books, menus, even such dry
information as reading bulletins on food production became
fascinating subjects to men who, a few months earlier, gave little
thought to the fine points of the culinary art or the intricacies of
agricultural statistics" (p. 250). But these men, when examined
by psychological instruments such as a free-association test, the

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 227

Rorschach test, and Rosenzweig's Picture Frustration Test, gave
almost no indication of enhanced food-related responses. The only
positive result was that they made significantly more uncommon
(idiosyncratic) responses to the food words of the free-association
test than did a group of nonhungry control subjects. The finding
that a greater total number of responses was given to the Rorschach
test after six months of starvation than during an initial control
period is consistent with the idea that the drive due to hunger
should enhance all responses. But similar increases were not found
in the case of other measures, and it may be that this was simply
a retest phenomenon, as Brozek et al. suggest.

Just why the psychological tests yielded negative results is un-
clear, but positive evidence for the effects of hunger on behavior
might have been obtained had different tests been used. It is
difficult, on intuitive grounds, to believe that either the Rorschach
test or the Rosenzweig Picture Frustration Test provide stimuli
capable of eliciting differential responses from hungry and non-
hungry subjects. This is indicated by the fact that seven subjects
never gave a food response to the Rorschach test, and only
eight of the 100 stimulus words used in the free-association
test were clearly related to food and eating. Perhaps, therefore, a
closer relationship between the psychological-test data and other
behavioral criteria would have been obtained had the tests con-
tained a larger proportion of hunger-related stimulus items.

The investigation of Brozek et al. involved a more protracted
and severe deprivation state than that of any other experimental
study. Short-term deprivation effects, however, have been studied
in a few instances. For example, Sanford (1936), in what may
have been the first attempt to investigate the influence of hunger
on imaginal processes, administered word-association and picture-
interpretation tests to school children both before and after lunch.
More than twice as many food-related responses were elicited by
the prelunch tests as by the postlunch tests. In a subsequent study,
using college students, Sanford (1937) found a progressively in-
creasing tendency for food-relevant responses to occur on a battery
of tests as deprivation was increased up to 24 hours.

In an experiment on food deprivation and imaginative processes,
Atkinson and McClelland (1948) asked subjects who had been

228 THE MOTIVATION OF BEHAVIOR

without food for 1, 4, and 16 hours to write short stories about
pictures drawn principally from Murray's Thematic Apperception
Test. The specific pictures were chosen to represent a variety of
aspects of situations related to hunger, such as satiation, depriva-
tion, food, and an eating place. The writers were asked to make
their stories as imaginative and interesting as possible, and the
results were scored by noting the percentage of writers whose stories
exhibited certain characteristics. These characteristics or categories
were, for example, story plots about food, statements about food
deprivation, about activity instrumental to the securing of food,
and about consummatory activity. Tabulations of responses in
these categories revealed that as hunger increased, there was no
over-all increase in the percentage of subjects who used food
imagery or food themes (plots) in their stories. Moreover, there
was a decided decrease in the number of references to the goal
activity of eating as a function of deprivation time. Hunger was
accompanied, however, by an increase in the percentage of stories
in which food-deprivation plots and deprivation-overcoming in-
strumental activity appeared. The over-all findings were thus am-
biguous, since the hypothesized relation between need and need-
related imaginative responses was supported by the data from some
scoring categories but not by the data from others.

The significance of the particular stimulus items used in tests
designed to evaluate the effects of deprivation on performance is
underscored by the results of the Postman-Crutchfield (1952) ex-
periment. These investigators varied not only degree of hunger
but also the probability that a stimulus would elicit a food re-
sponse as well as the strength of the subject's set to make food
responses. The stimuli were skeleton words, with blank spaces to
be filled in by the subjects. Different types of lists having low,
moderate, and high probabilities of eliciting food responses were
used in conjunction with levels of hunger ranging from 0 to 6
hours. The degree of selective set to give food responses was varied
by arranging the first few words of the lists so that 0, 1, 2, or 5 food
responses would be elicited from different groups of subjects prior
to their being presented with the principal items of the list.

Analyses of data from rather large numbers of subjects showed
that when degree of food set and type of list (food-response prob-

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 229

ability) were held constant, no simple relation existed between
intensity of need and frequency of need-related responses. Degree
of set, however, as induced by increasing numbers of prelist
food responses did lead to progressively more frequent food solu-
tions to the skeleton words. The only way in which hunger affected
performance was to produce slight changes in the relative effec-
tiveness of the different levels of set. That is, for hungry subjects
(4 to 6 hours of deprivation) the function relating degree of set
to number of food responses was positively accelerated, whereas
for nonhungry individuals (0 to 1 hour) the function was nega-
tively accelerated. It was also found that the effect of set on fre-
quency of word solutions was greater for skeleton items having a
moderate probability of eliciting food words than for items with
either a very low or a very high probability. Nonselective energiz-
ing effects of hunger could not, of course, be revealed by this ex-
periment since every subject was required to complete the same
total number of skeleton words.

Generally speaking, it is clear from the Postman-Crutchfield
study that the selective effects of mild degrees of deprivation are
slight and can be demonstrated only by carefully designed experi-
ments in which a delicate balance is maintained among such vari-
ables as set and initial response probability. Quite certainly, if all
skeleton words elicited food responses, or if none did, it would be
impossible to demonstrate that variously deprived groups of sub-
jects differ with respect to food-response output. The test items
must be carefully chosen if group or individual differences are not
to be obscured by either "ceiling" or "floor" effects. This is not,
incidentally, a problem unique to the measurement of deprivation
effects. It is characteristic of all measurement situations. It would
be impossible, for example, to find out how well children solve
arithmetic problems if the problems are either so easy that all
children obtain scores of 100, or so difficult that none can be solved
by any student.

In only one experiment, apparently, has the relation of hunger
to eyelid conditioning been explored. Franks (1957) studied the
acquisition and extinction of eyelid responses in nondeprived sub-
jects and in subjects who had gone without food, water, and
tobacco for approximately 22 hours. Since the two groups did not

230 THE MOTIVATION OF BEHAVIOR

differ significantly in number of conditioned responses, the results
failed to support the predictions of a multiplicative-drive theory.
However, the procedures followed by Franks (massed acquisition
trials, partial reinforcement, few reinforced trials) were rather un-
conventional, and the asymptotic conditioning levels for both
groups were unusually low. A replication of this experiment under
conditions more favorable to the development of conditioned re-
sponses is clearly indicated.

The possibility that hunger might selectively affect the reten-
tion of previously learned materials has been investigated by Wil-
liams (1950). In his studies hunger, which was introduced after
original learning had been completed, had no effect upon the
retention either of words that were directly related to the need
or of neutral words that had been associated with need-related
words.

By way of summary we are led to conclude that exceptionally
little is known of the effects of deprivation on human performance
in nonperceptual tasks. We do not know, for example, whether
hunger tends to facilitate a broad diversity of responses, as a mul-
tiplicative-drive theory would predict, since most investigations
have not been designed to reveal such effects. Nor can we assert
with confidence that hunger always functions selectively. More-
over, experiments involving deprivation conditions other than hun-
ger are practically nonexistent, and we possess neither reliable
data on the learning of various tasks under different drive levels
nor information on the effects of a reduction in a primary need
following the performance of successful responses. The sum and
substance of our current knowledge seems to be that under rather
restricted conditions hunger sometimes exerts a mildly facilitative
effect upon the tendency to give verbal responses related in one
way or another to food and to the activities connected with its
procurement and consumption. Under other experimental circum-
stances, however, no relation between hunger and performance
has been observed.

Motivating Effects of Strong Stimuli

With human beings, as with other organisms, it has long been
supposed that the intense stimulation provided by electric shocks.

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 231

bright lights, loud noises, blasts of air, and internal tensions may
function as a motivational variable.

Electric Shock. Probably because of the ease with which it can
be administered, electric shock has been used with a wide variety
of organisms and in almost every conceivable laboratory situation.
In spite of its widespread use, however, standardized techniques
are still not available for administering and measuring shock, and
the results of different experiments, even when they purport to
be investigating the same problem, are seldom in agreement.

Overt changes in behavior produced by electric shock are quite
diverse, and depend not only on shock intensity but also upon its
time of presentation with respect to a response or response se-
quence. Obviously shock can be administered consistently either
before, during, or after a response has been elicited by other
stimuli. Or it can be introduced at random intervals bearing no
systematic temporal relation to any specific response. Let us con-
sider these two modes of shock presentation in turn.

If one were to try to simulate, with electric shock, the motivat-
ing characteristics of hunger, it would probably be necessary to use
a relatively mild continuous shock which would become progres-
sively stronger over a period of hours. One might then study
behavior in standardized situations with different initial and/or
terminal levels of shock and with partial or complete shock reduc-
tion following the performance of specific responses. But since
most human subjects object to such treatment, experimenters have
typically used rather brief shocks, the interval between response
and shock being held constant. On theoretical grounds we would
anticipate that shock administered following a response would
have a different effect on behavior than would shock presented
while a response is in progress or prior to its appearance.

The consistent administration of shock following a particular
response presents us with the typical paradigm of the punishment
situation, and the observation that response strength is thereby
weakened supports the belief that shock is functioning as a moti-
vational variable, since one criterion (Chapter 2) for the identifi-
cation of such variables is thereby met. The literature in this area
is voluminous, but the experimental results are by no means uni-
form. Probably the most reasonable general conclusion one can
draw from these studies is that behavior consistently preceding

232 THE MOTIVATION OF BEHAVIOR

very strong shocks tends to be abandoned (cf. Postman, 1947),
but there are numerous exceptions to this rule.

The weakening of a response as a consequence of subsequent
punishment can be interpreted in several ways. For example, if
shock regularly follows a response, the environmental cues eliciting
that reaction, as well as internal response-produced cues, should
acquire the capacity to evoke conditioned fear reactions (cf. Chap-
ter 5 ) . Initially, fear-produced increments in drive might augment
the strength of the punished response, but subsequently, with
extremely strong fear, competing responses might be aroused and
their appearance in lieu of the punished response might be rein-
forced by a reduction in fear. Alternatively, postresponse shocks
could elicit escape or withdrawal reactions that might be reinforced
by shock cessation and might competitively interfere with the prin-
cipal to-be-punished reaction. For instance, a subject may be
shocked on the fingers each time he enters a blind alley in a finger
maze. If the shock is terminated whenever he moves his hand
away from the end of the alley, such escape movements may be
reinforced; and since these newly acquired movements are op-
positely directed to those involved in entering the blind, the latter
may be overridden or weakened.

The administration of rather mild shocks after the occurrence
of particular responses may serve simply to inform the subject as
to the correctness or incorrectness of his reactions. Thus it has
been shown, by Tolman, Hall, and Bretnall (1932), and by Muen-
zinger (1934), for example, that shocks following correct responses
in a maze are sometimes as effective in producing learning as are
shocks following errors. Shocks administered in this way may or
may not have significant motivational effects, but they clearly
qualify as associative variables since they provide specific knowledge
of results. Weak shocks probably lead neither to escape reactions
nor to increments in drive, and conditioned emotional reactions
should seldom develop from their being paired with neutral cues.

The introduction of shock prior to the occurrence of a response
might be expected to have motivational consequences if shock-
induced emotionality or other processes persist until the response
is evoked. Or if shock precedes a response by only a very brief
interval, shock-elicited skeletal responses might perseverate for a

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 233

time and interfere with the subsequently ehcited criterion reaction.
In this case, of course, the dechne in response strength would
qualify as an instance of competitive interaction among responses
and hence as an example of an associative effect.

Precisely similar analyses would appear to be applicable to
situations in which shock is administered coincidentally with an
ongoing reaction. On a multiplicative-drive theory any shock
should intensify whatever behavior is in progress and thereby fulfill
the energizing criterion of motivational variables. But the more
intense the shock, the higher the probability that it will elicit
strenuous responses of vocalization or escape and thereby lead to
a decline in the efficiency of overt performance. In every instance
shock can apparently function both as a source of drive and as a
specific stimulus capable of arousing a variety of defensive reac-
tions. When behavior is facilitated by shock, the effect can be
ascribed either to an increase in drive (motivational theory) or
to the appearance of a facilitative response (associative theory).
Likewise, performance degradation can be attributed to the ap-
pearance of shock-induced competing reactions, or to an initial
superiority of task-irrelevant reaction tendencies in comparison
with the correct ones.

If electric shocks are presented at random intervals they cannot
provide reliable task-specific information to the subject nor is it
likely that stable, stimulus-specific conditioned fear reactions will
develop. Nonetheless, behavior might be affected either positively
or negatively, since the subject's general drive level might be in-
creased and/or competing responses might be evoked. Experi-
mental evidence indicating that random shocks sometimes lead to
improved performance has been reported by several investigators
(e.g., Gilbert, 1936, 1937; Rosenbaum, 1953; Spence, Farber, and
Taylor, 1954), but such shocks also have been found to disrupt
the performance of some subjects (Deese, Lazarus, and Keenan,
1953). In so far as random shocks function facilitatively, support
is provided for the multiplicative-drive view, since these shocks
do not provide cues appropriate to the solution of a problem.

Unconditioned Stimuli as Sources of Drive. In the experiments
we have just reviewed, the reference responses being studied were
typically not elicited by shock. However, when shock is used as

234 THE MOTIVATION OF BEHAVIOR

the UCS in human conditioning studies, it probably functions both
as an ehcitor of the to-be-conditioned reaction and as a source of
drive. In a typical experiment (e.g., eyelid conditioning), if a
response is to be counted as "conditioned," it must occur after
the onset of the CS and, generally speaking, before the onset of
the UCS. Consequently, on any one conditioning trial, the drive
produced by the UCS cannot affect the latency, and perhaps not
even the magnitude, of a foregoing, anticipatory CR. Both latency
and amplitude could be affected, however, by residual drive from
UCS presentations on earlier trials. Thus, as Spence (1958) has
surmised, if one group of subjects is conditioned v^ith a strong air
puff and another with a weak air puff, the former should have a
higher average drive level than the latter. Because of this the fre-
quency of CRs should be directly related to the intensity of the
puff, or more generally, to UCS intensity.

Available experimental data, though surprisingly sparse, support
these expectations. In what may have been the first study of this
kind, Passey (1948) reports that mean frequency of conditioned
eyelid responses increases in an approximately linear manner as
a function of the logarithm of air-puff intensity. Subsequently,
Spence (1956, 1958) and his associates have repeatedly shown
that conditioned eyelid responses are more frequent when strong
rather than weak puffs are used. The dashed-line curve of Fig. 7:4,
which appears in another context later on in this chapter, clearly
shows this effect. In addition, acquisition curves obtained with
different UCS intensities tend to diverge during the course of
training. This finding is consistent with the assumption that the
habit strengths for the groups under comparison are multiplied
by different magnitudes of drive (D). Also relevant here is Mc-
Allister's (1953) study in which it was shown that the eyebhnk
response is extinguished far more rapidly when the UCS is omitted
entirely, and hence drive level is lowered, than when the UCS
continues to be presented, though separated from the CS by an
interval too long to produce conditioning.

Unfortunately, no UCSs other than air puffs have apparently
been systematically manipulated in classical conditioning experi-
ments with human subjects. It remains to be seen, therefore,
whether the consistent results obtained in the eyelid situation
can be reproduced with other aversive stimuli and other responses.

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 235

It is also evident that other interpretations are possible beside that
which attributes drive-arousing properties to the UCS. Variations
in UCS level might affect performance not only by changing drive,
but also by altering habit strength or by evoking different classes
of unconditioned responses.

Noise. Industrial psychologists have long been concerned with
the problem of how a worker's efEciency is affected by the loud
noises that often accompany manufacturing processes. Experi-
mental attempts to investigate this problem have not, however,
pointed unambiguously to any one general conclusion. In some
cases noise seems to have a facilitative, dynamogenic effect on
performance, though marked increases in energy output and in
fatigue are typical accompaniments of such effects. In other cases,
especially where the noise, because of its intermittency or other
special characteristics, appears to elicit competing reactions, per-
formance declines. By and large, however, the most extensive
studies of performance under noise, such as those of Stevens and
his collaborators (1946), indicate that performance on a variety
of psychomotor and intelligence tests is not significantly changed
by prolonged exposure to noise. This may or may not mean that
noise has no motivational effects. Quite possibly, noises of certain
kinds may produce increments in drive strength, and yet, at the
same time, elicit interfering responses. To date there are few
experiments on performance under noise in which adequate num-
bers of subjects have been tested and in which irrelevant factors
of suggestion and the like have been adequately controlled (cf.
Chapanis, Garner, and Morgan, 1949). We must conclude, there-
fore, that while noise may function as a motivational variable, its
effects are in marked degree a function of its specific characteristics,
especially as these relate to the rhythm or other aspects of the
task, to the operator's level of proficiency, and to numerous other
variables.

Motivating Effects of Instructions

Verbal instructions, whether administered before, during, or
after a bit of behavior has been exhibited, are stimuli, which, like
electric shock and noise, may have motivational consequences. As
we have already remarked in discussing acquired sources of drive,

236 THE MOTIVATION OF BEHAVIOR

nonspecific "hurry-up" instructions seem to function motivationally
to facilitate whatever behavior is taking place or is about to take
place. Besides, individuals seem to learn to motivate their own
activities by administering similar "try-hard" instructions to them-
selves. The laboratory psychologist is well aware that performance
levels can be changed if subjects are told that the task is impor-
tant (or unimportant), that the test is a measure of intelligence,
that the results will affect their standing in class, or that they have
failed or succeeded. As with all variables alleged to have motiva-
tional effects, however, instructions may modify performance either
by virtue of their cue properties and the resulting changes in asso-
ciative strengths or through a change in drive strength, or in both
ways.

When verbal instructions contain specific cues appropriate to
the responses being scored as correct, improvement in performance
is expected and is seldom considered to be a motivational outcome.
For example, if a subject were told which way to go at each choice
point of a maze in order to stay on the true path, his improved
performance in comparison with that of noninstructed subjects
would not qualify as a motivational effect. But if he were simply
told to do his best, and if this instruction were followed by superior
performance, we would be more likely to conclude that a change
in drive had been produced. Instructions to "do your best" provide
no specific response-directing cues, and if improved performance
follows it may be due to nonassociative processes. This is not to
say, however, that a nonmotivational interpretation of such results
could not be defended. One could always argue, for instance, that
alleged motivation-inducing instructions augment performance not
because they add an increment to drive, but because they arouse
a whole family of learned facilitative responses such as sitting
erect, concentrating on the task at hand, and suppressing inclina-
tions to daydream.

Instructions to subjects to the effect that they have either failed
or succeeded in the performance of an assigned task are commonly
held to exert motivational influences on subsequent tasks. But
here again, as Lazarus, Deese, and Osier (1952) and Farber (1955)
have observed, it is desirable to distinguish between the purely
motivational effects of such instructions and their associative (in-
formative) consequences. For the most part, the kinds of instruc-

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 237

tions described by the terms success and failure tend to be unre-
lated to the particular responses involved in performing a just-
completed task. They tend, that is, to be relatively general state-
ments such as "You have done better (or worse) than most people
do on this task." The nonspecific nature of such instructions
minimizes, though it does not eliminate, the likelihood that sub-
sequent behavior will be affected through associative channels. The
most general finding of studies of instruction-induced failure seems
to be that subsequent performance is impaired (cf. Farber, 1955),
but this effect could be interpreted equally well by either a motiva-
tional or an associative theory. As Farber notes, "... the impair-
ment of performance by failure does not necessarily demonstrate
its associative effects, nor does the improvement of performance
by failure demonstrate its drive effects. Just as the associative mech-
anism attending failure may either benefit or harm performance,
so might an increase in drive, in and of itself, affect behavior
either favorably or adversely, depending on the specific nature of
the task and the experimental conditions involved" (p. 317).

Because of the conflicting nature of experiments designed to
manipulate human motivational level by means of verbal instruc-
tions, and because none seems to have been designed so that the
separate contributions of drive and associations can be evaluated,
we have chosen not to review them in detail. Praise or reproof
may lead either to better or poorer performance, depending upon
the nature of the experimental conditions, which are as yet poorly
understood, and upon the characteristic ways in which individual
subjects react to those conditions, to the experimenter, and to
his comments concerning the subject's performance. About all one
can say with assurance is that associative factors probably play a
dominant role when the instructions are related to the correctness
or incorrectness of specific responses and that a concept of a
nonspecific drive may be useful when success-failure statements
are very general.

Response-defined Motivational Level

In the preceding sections of this chapter we have dealt briefly
with some theoretical and experimental aspects of human mo-
tivational differences as defined by amount of deprivation and

238 THE MOTIVATION OF BEHAVIOR

intensity of stimulation. The remainder of the chapter is devoted
to studies in which degree of drive or motivation is defined, not by
the manipulation of an external variable, but in terms of individual
differences in responses to standardized tests. In the studies to be
considered, motivation is said to differ among individuals who
respond differently to the stimuli provided by projective or ques-
tionnaire tests.

As the reader may recall from Chapter 2, drive level can be
operationally defined, independently of the behavior to be ex-
plained, by noting how different subjects react in a given situation.
The finding of stable individual differences in reactivity permits
us to rank-order our subjects with respect to test performance, and
if we choose to define drive in terms of test scores, we can then
rank-order our subjects with respect to drive. Drive differences thus
defined may never, of course, turn out to be significantly related
to performance in any other situation or to other theoretical con-
cepts. If not, then other ways of defining drive may be tried and
the most useful definition provisionally adopted.

Psychologists have sometimes claimed that the use of response-
defined drive levels is to be avoided as not "truly experimental,"
since drive is not being manipulated directly. That this objection
is unfounded can be made clear by the following example. Suppose
we desire to vary the intensity of a stimulus light in our laboratory.
We can achieve this variation either by adjusting the amount of
electric current that is allowed to flow through the filament of a
single lamp, or by selecting several different lamps from an avail-
able population of lamps. The first method parallels that in which
drive is "manipulated" by, say, changing the strength of shock
administered prior to a behavioral test. The second is like that in
which individuals having specific degrees of the property in ques-
tion, i.e., strength of drive, are chosen from a large population
by means of a test. Since standardized instruments for measuring
light intensity have been developed by the physicist, the readings
of such instruments can be used to support the contention that
selective and manipulative procedures can lead to the same out-
come. In principle, the selective and manipulative methods of
varying drive can also be equated, but until practical instruments
for assessing drive strength have been developed and standard-

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 239

ized, any assertion to the effect that a particular level of test-de-
fined drive is identical with one produced experimentally must
remain open to question.

Critics of the selective procedure have argued further that
individuals who differ with respect to their responses on an al-
leged test of drive may also differ in a variety of other ways, thus
confounding other variables with drive. The comment, though
legitimate, is not a criticism uniquely relevant to the selective pro-
cedure. It applies with equal cogency to the manipulative methods
as well. Thus when we try to change drive by experimentally alter-
ing deprivation time or by administering different strengths of
shock, we may also be altering associative strengths, hormonal
levels, and many other factors. Likewise, in manipulating the
intensity of light from a single bulb, changes in the color of the
light may be produced, just as bulb diameter may fluctuate when
intensity is varied by the selection of particular lamps. However,
if our standard light-measuring device is insensitive to color and
to bulb diameter, then the presence of these confounding variables
is of no consequence. And in the case of drive, if there is reason
to believe that the behavior being studied is insensitive to the pres-
ence of known or suspected confounding variables, then changes
in such ancillary factors, whether produced by manipulative or
selective methods, can be ignored. In principle, therefore, there is
as much justification for trying to vary drive by a process of selec-
tion as by seemingly more direct experimental manipulations. No
one can predict, at present, which of the many varieties of each
method will ultimately yield the most useful and significant esti-
mate of drive, nor whether one of the methods will ultimately
prove to be better than the other.

Motivation Defined by Imaginative Responses

One of the most extensive research programs directed toward
evaluating human motivation by means of subjects' responses is
that of McClelland and his principal associates, Atkinson, Clark,
and Lowell (1953).

As we have already seen (Chapter 5), these authors use the
word motive to designate learned, and only learned, anticipations

240 THE MOTIVATION OF BEHAVIOR

(expectations) of rewards or punishments. An individual has a fear
motive or an anxiety motive if, when exposed to cues that have
previously been followed by punishment, he experiences an affec-
tive (emotional) arousal similar to, or representative of, the arousal
produced by pain or punishment. An organism has a hunger
motive when, as a consequence of repeated eating experiences, the
internal cues accompanying deprivation or external cues (or both)
arouse an affectively toned expectancy of the pleasurable conse-
quences of eating. We have previously noted that these negative
and positive motives are essentially identical, in their origins and
affective aspects, with conditioned anticipatory fear reactions and
anticipatory fractional goal responses, respectively. The achieve-
ment motive, which has been studied most extensively, is defined
as a learned, affective anticipation of the pleasurable consequences
of success in situations where the quality or efficiency of one's
performance is to be evaluated. Overt behavior is assumed to be
affected "motivationally," though in ways that are never clearly
specified, by the arousal of one or more motives such as these.

Measuring Human Motive Strength. In attempting to develop
an adequate definition of human motive strength, McClelland
and his coworkers have concentrated their attention upon the
content of imaginative responses given by subjects to a standardized
test. Specifically, subjects are presented with a series of relatively
uninformative pictures resembling those in the Thematic Apper-
ception Test, and are asked to tell an imaginary story about each.
The stories are then scored by counting, according to certain rules,
the number of ideational items believed to be diagnostic of the
presence of a specific motive. Individuals or groups can thus be
rank-ordered with respect to the strength of a given motive as
defined by the frequency of their motive-relevant reactions to the
pictures. The performance of the same individuals under other
conditions can then be studied to see whether it is meaningfully
related to fantasy-defined motive strength.

This technique of defining individual differences in motive
strength in terms of subjects' reactions to one situation (story
telling) and of using such differences to explain performance
variation in a second situation exemplifies the response-based
method (Chapter 2) of independently defining drive. The proce-

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 241

dure entails the working out of an R-R relation, and involves the
assumption that the motive governs both imaginative reactions
and test-situation performance.

In principle, any set of reactions to a standardized test can be
used to define motive strength, but McClelland et al. regard the
fantasy content of stories as uniquely suitable. Their preference
for this method rests upon the traditional psychoanalytic view
that unguarded free associations provide the richest material for
studying human motives, upon their belief in the fruitfulness of
the method, and upon the questionable assumption that fantasies
are not affected by a subject's specific knowledge or factual in-
formation.

Experimentally Aroused Motive Strength. One of the first ex-
periments in this area, one we have already briefly discussed, was
performed by Atkinson and McClelland (1948). These investiga-
tors reasoned that if the content of imaginative thought reflects
motive strength, changes in that content should follow from ex-
perimentally induced variations in the intensity of a motive. An
adult human being who has been deprived of food should possess
a hunger motive, and the presence and strength of this motive
should be revealed through imaginative stories by an increase in
the frequency of food- and/or eating-related items. As we have
seen, Atkinson and McClelland's expectations were not borne out
in the case of food-related responses, though increased hunger was
accompanied by more frequent references to food deprivation.
Nevertheless, this minimally satisfying relation was taken as con-
firming the hypothesis that with lengthened deprivation the hun-
ger motive was aroused more vigorously by the combined action
of deprivation and picture cues, and that the intensified motive
enhanced the probability of occurrence of food-related imagina-
tive responses.

By means of additional experiments McClelland and his col-
laborators have tried to show that conditions designed to induce
the arousal of the motives of sex, fear, affiliation, and achievement
also lead to corresponding increases in the motive-related content
of imaginative stories. With the exception of Clark's (1952) ex-
periment on the sex motive, stories produced under conditions
regarded as motivating ". . . contained more imaginative responses

242 THE MOTIVATION OF BEHAVIOR

dealing with thoughts, feehngs, and actions related to the goal-
directed sequence of the motive in question" (Atkinson, 1954,
p. 66). For reasons not yet well understood, however, these results,
obtained with male subjects, could not be replicated with women,
Wlien the same procedures were tried with female subjects, as
many motive-related imaginative responses were obtained when
the motives were supposedly aroused as in nonaroused control con-
ditions.

With these experiments as a foundation, the conclusion was
reached that the strength of almost any motive could be estimated
from the frequency with which certain types of imagery appeared
in subjects' stories about pictures. Thus if one person gives more
food-related responses to pictures suggesting eating than does
another, the inference is drawn that the former is hungrier, has a
stronger motive to eat, than the latter. And the person whose
imaginative stories contain frequent references to success and
achievement is presumed to have a stronger motive to achieve
than one whose imaginative output is less rich in words connoting
goal attainment.

The Achievement Motive and Overt Behavior. Much of the
research effort of the McClelland group has been directed toward
determining the ways in which motives, especially the need for
achievement, influence overt behavior other than that involved in
the telling of imaginative stories. They have sought to determine
whether groups of individuals defined as having strong need for
achievement will perform more effectively on various tests or tasks
than low-achievement groups.

That need for achievement apparently does not function as a
nonspecific energizer after the manner of Hull's D is suggested by
the data from several investigations. Thus Zatzkis (1949) failed
to find a clear or significant relation between the mean number
of words written in a class essay and the need-achievement scores
of the student writers. Moreover, McClelland et al. (1953) re-
ported that subjects with low need-achievement scores gave a
greater mean number of responses to the 10 cards of the Rorschach
test than did the high scorers, with moderate need-achievers giving
more responses than either high or low. And finally, Atkinson
(1950), in a study preparatory to one on the recall of completed

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE

243

and uncompleted tasks, found that with a Hmited number of
tasks subjects who scored high on the need-achievement test never
completed significantly more tasks than did subjects with low
need-achievement scores. In fact, under instructions designed
to be relaxing rather than motivating, the high need-achievers per-
formed significantly fewer tasks than the low need-achievers.

Inasmuch as McClelland and his collaborators have never speci-
fied precisely what it is that a motive does when functioning as a
motivator, considerable uncertainty remains as to how overt be-
havior should be affected by a motive such as the need for achieve-
ment. Generally speaking, however, these investigators appear to
hold that increased motive strength should lead to faster perform-
ance and greater work output, or, under unspecified conditions,
to more efficient performance.

Perhaps the best support for the expectation that high need-
achievers should perform better than those with low scores is
provided by Lowell's (1952Z?) experiment. His college student sub-
jects were required to perform a simple arithmetic task for a period
of ten minutes. Tabulations of the number of addition problems
solved during each two-minute period showed that subjects high
on the achievement scale performed more efficiently during each
practice period than subjects who were low on the scale. These
data, which are presented graphically in Fig. 7:1, are cited by
Atkinson (1954) as the kind of relation that "has often been used

Fig. 7:1. Performance
levels obtained on a sim-
ple addition task by in-
dividuals v^fho had at-
tained high and low
scores on a need-for-
achievement test. {From
Lowell, 19S2b.)

'?;28

High need-achievement
A' = 19

Low need-achievement
N = 2l

S y'

.T.

2 3 4

Two-minute periods

244

THE MOTIVATION OF BEHAVIOR

to illustrate the so-called 'energizing' function of motives" (p. 71).

Of special interest in connection with this experiment is the
reported fact that the subjects of the two groups did not differ
significantly in respect to their basic numerical skills as measured
by the quantitative score on the ACE Psychological Examination.
This would seem to imply that achievement motivation was func-
tioning as a nonselective energizer to enhance performance, since
it indicates that the basic skills (associative factors) of the two
groups were equal. But from another point of view this is puzzling.
On the basis of published statements concerning the nature of the
achievement motive, one would expect the motive to be aroused
during the taking of the ACE test to an even greater degree than
during an experiment requiring the performance of simple addi-
tions. Consequently, it may be appropriate to ask why the high
need-achievement group differed from the low group on the simple
addition tests but not on the quantitative portion of the intelli-
gence test. Performance on an intelligence test involves the efficient
utilization of knowledge and conceptual skills as well as basic
capacity, and highly motivated individuals should perhaps perform
more efficiently on such tests.

In another part of Lowell's experiment, high and low need-
achievers were required to reconstruct common words from scram-
bled arrangements of the letters. On this task, as Fig. 7:2 shows,

High need-
achievement

-Low need-
achievement

3 4 5 6 7

Two-minute periods

Fig. 7:2. These data indicate that subjects with high need-achievement scores
tend to improve during the course of an anagrams task, whereas subjects with
low scores do not. (Adapted from Lowell, 1952b.)

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 245

the two groups did not differ greatly at first, but the high need-
achievement group showed a progressive increase in facihty with
practice while the other group did not. The difference between
the over-all means for the two groups was not statistically signifi-
cant, but the divergence of the two curves was said to be highly
reliable. This finding has been interpreted (McClelland et al.,
1953) to mean that where learning is possible, as in this scrambled-
words task, subjects with strong achievement motivation learn as
the task progresses, and the skills thus acquired are facilitated by
the achievement motive. Moreover, the occurrence of learning is
said to strengthen the view that need-achievement scores do indeed
reflect level of motivation. The view is based on the supposition
that the mere occurrence of learning constitutes a decisive criterion
for evaluating the presence of a motive. Here the authors have
failed to see that for most theorists learning provides an adequate
criterion of motivation only when a reduction in motivation has
clearly been involved.

Lowell's finding that the so-called highly motivated group per-
formed more poorly than the low group during the first two min-
utes of the test is also rather puzzling. Assuming equal skills in
unscrambling the mixed-up words, we should expect a more highly
motivated group to do better initially than a poorly motivated
one. The failure to find an initial difference is even more disturb-
ing when we realize that relevant skills were probably not equal
at the outset, since the group with high achievement scores was
significantly superior to the low group on the linguistic score of
the ACE test. The divergence of the curves probably can't be
explained, however, in terms of differences in linguistic ability
alone. When this "intelligence" factor was held constant by the
technique of partial correlation, there was still a significant correla-
tion (r = .44) between achievement motivation and output of
scrambled words.

Additional evidence purporting to show that performance in
certain situations is enhanced by high achievement motivation
has been reported by McClelland and Liberman (1949). Accord-
ing to these investigators, subjects with high achievement motiva-
tion exhibit lower recognition thresholds for words denoting sue-

246 THE MOTIVATION OF BEHAVIOR

cess than do subjects with low achievement motivation. This may
or may not be a motivational phenomenon, however, since the two
groups might not have been equally familiar with the critical
stimulus words. Moreover, Veroff, Wilcox, and Atkinson (1953)
showed that women with high achievement-motivation scores did
better than those with low scores during the middle portion of an
anagrams test when the general level of performance tended to lag.
Hedlund (1953), however, was unable to replicate this finding.
Evidently we cannot yet accept the conclusion that performance
in a variety of situations is enhanced by high achievement mo-
tivation. Furthermore, as Atkinson (1954) has noted, a number
of the supportive experiments involve behavioral measures con-
cerning which the theoretical expectations are not especially clear.
How Motives Affect Behavior. In their early publications Mc-
Clelland and his associates failed to indicate how motives, as
defined by them, function as determinants of verbal fantasy re-
sponses or of responses in problem solving and other task situations.
Atkinson (1954) has tried to fill this gap by relating the concept
of motives to the expectancy theories of MacCorquodale and
Meehl (1953) and of Tolman and Postman (1954). Considering
first the question of how motives might affect imaginative responses
to TAT-like pictures, it is clear that here Atkinson has adopted
a straightforward associative view. Thus he maintains that while
motives are being learned many additional reactive tendencies
(response predispositions) are also becoming associated with the
same cues. The internal stimuli of hunger, for example, tend to
arouse not only the anticipation of food (the hunger motive) but
also verbal and other responses that are related to eating, to a
wide variety of different foods, and to the situations in which
eating has frequently taken place. Consequently, when pictures
containing food-related cues are presented to hungry subjects, the
combination of internal and picture-provided cues tends to elicit
food-related imaginative responses. In essence, the stronger the
hunger motive, the stronger the learned predisposition to use words
related to eating, and the greater the likelihood that such words
will be evoked by the joint action of picture cues and the com-
bination of hunger-produced and anticipation-of-food-produced
cues. Similarly, a high achievement motive should be accompanied

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 247

by strong predispositions to perform certain kinds of responses
(including words), and in telling imaginative stories, therefore,
subjects so motivated should be more likely to use words related
to achievement. It is thus evident that a motive, defined as a
learned, affective, anticipatory process does not function, with
respect to fantasy responses, as a drivelike energizer, but only as
a source of stimuli to which specific kinds of associations have
been developed. Only the single principle of interactions among
families of associative tendencies is involved in the theoretical
explication of the alleged motivationlike effects of motives upon
imaginative verbal responses.

In dealing with behavior in problem-solving and learning situa-
tions, Atkinson appeals to two different kinds of expectancies.
One of these, the motive, is the affectively toned expectation of
either good or bad things to come. The second kind of expectancy
is described as cognitive or perceptual rather than affective. Such
an expectancy is a consequence of the organism's repeated expe-
riences with environmental objects and their interrelations. It is
approximately equivalent to Tolman's (1948) notion of a "cogni-
tive map" and to Hull's construct of habit strength. This cognitive
expectancy, much like Hull's habit strength, is apparently in-
capable, when acting in isolation, of leading to overt behavior. A
person may know that food can be obtained from a neighborhood
store and precisely how to get there (cognitive expectancy), but
he will not travel to the store to get food unless he has a desire
to eat (motive-type expectancy). According to Atkinson, the final
tendency to respond overtly depends jointly upon these two kinds
of expectancies.

Atkinson's treatment still leaves us in doubt, however, as to the
"motivating" role of the motive-type expectancy. In one of his
diagrams the motive is shown as multiplying the environmentally
cued performance expectancy, but in the accompanying text we
are told only that "mutual facilitation" increases the strength of
the disposition to respond. Apparently, the use of the multiplicative
sign in this diagram was not intended to parallel Hull's use of
the same symbol, since Atkinson holds that a motive facilitates
only those instrumental reactive tendencies that have a terminal
member, the goal concept, in common with the motive. This sort

248 THE MOTIVATION OF BEHAVIOR

of selective enhancement is evidently more representative of an
associative than of a motivational interpretation. On this view, if
an animal is made hungry and is then placed into an alley where
it has previously received only water when thirsty, the two ex-
pectancies would not have a common goal member. The environ-
mental cues would arouse the expectation of obtaining water in
the goal box, but the hunger motive would arouse the anticipation
of eating food. Hence, this theory would seem to predict that the
hungry animal's instrumental responses of running down the alley
and especially his responses of drinking water would not be en-
hanced by the presence of the hunger motive. However, as we have
seen in our review of Miller's (1948b) experiments (Chapter 3),
hunger does seem to facilitate responses previously learned under
thirst.

In summary, it appears that a motive, as defined by McClelland
and his associates, always functions as a selective facilitator, capa-
ble of enhancing only restricted sets or classes of responses. In this
respect a motive differs markedly from a motivational concept
such as Hull's drive, which is assumed to multiply all learned and
unlearned reactive tendencies. The two conceptions also differ, in-
cidentally, in that Hull's theory predicts that increased drive will
lead to poorer performance under certain conditions, whereas the
motive theory always seems to predict enhanced performance. In
addition, the mechanisms through which motives exert their selec-
tive effects either upon imaginative responses or upon behavior
in learning and problem-solving situations seem to involve no prin-
ciples other than those of conventional associative learning theory.
This may be a virtue, but it has naturally led certain critics (e.g.,
Farber, 1954b) to wonder whether the term motivational is,
strictly speaking, appropriate to either the theory or the research
of McClelland and his coworkers.

Selective Response Facilitation. Tlie view that a motive func-
tions selectively to enhance motive-related responses presents an
interesting problem for theoretical analysis. Let us suppose we
were to perform an experiment like that of Atkinson and McClel-
land (1948), and that our subjects tended, as time of food depriva-
tion increased, to give greater numbers of imaginative responses
related to food and eating. If our pictures had been appropriately

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 249

chosen, the cues they provided might have aroused learned asso-
ciative tendencies to emit food responses which were equal in
strength to the tendencies to give nonfood reactions. If hunger
were functioning as a generalized drive (D), the excitatory tend-
encies to make both food-related and non-food-related responses
should be strengthened equally as deprivation time increased.
This would mean that although the proportion of food-related to
non-food-related reactions would be the same for all our groups,
the total number of responses of each kind would increase with
hunger. Hence, if between-group comparisons were made on the
basis of food-related words only, hunger would appear to have a
selectively energizing effect, when in fact it had not. On the
other hand, if our pictures elicited more food-relevant than non-
food-relevant words, we should expect, from a multiplicative-drive
theory, that the absolute difference between the number of food
and nonfood responses would increase across groups with intensi-
fied hunger. In this case, increases both in the absolute number of
food-related words and in their frequency compared with non-
food-related words would be predicted as deprivation was length-
ened. Apparently then, what seems to be a selective energizing
effect can, in theory, be produced by an increase in generalized
drive level. In one case this results from a comparison between
groups when deprivation-relevant and deprivation-irrelevant re-
sponse tendencies are equally strong, and in the other, from com-
parisons either between or within groups when deprivation-related
tendencies are initially dominant.

Selective sensitization of food-related responses could also be
produced, however, by the presence of specific associative tend-
encies established through prior learning to the cues accompanying
hunger (cf. Brown, 1953a). Thus intensified hunger should lead to
more food responses, provided food responses have been associated
with the internal stimuli of deprivation. Were such a mechanism
as this operating effectively, one would predict that the difference
favoring food-related over non-food-related words would increase
either within a single group or across groups as deprivation was
extended. Moreover, the effect should be obtainable even when
the external cues provided by test pictures tend to elicit initially as
many food-related as non-food-related responses. Obviously this is

250 THE MOTIVATION OF BEHAVIOR

an associative view, since selective sensitization is explained with-
out invoking either a multiplicative drive or a motive.

One might maintain, of course, that increased deprivation leads,
as in the above interpretation, to stronger internal cues and hence
to stronger habits but that these habits are reflected in overt be-
havior via their effect upon other habits. Thus the tendencies to
make food-related responses that are associated with deprivation
stimuli might combine with the reactive tendencies incited by
picture cues to increase relevant response frequency. This might be
treated conceptually as a matter of stimulus summation or in
terms of the facilitative effects of one habit set upon another. In
either event, the interpretation, which closely parallels that pro-
posed by Atkinson, would once again be associative rather than
motivational.

Alternative Proposals for Measuring Achievement Motivation.
According to McClelland (1956), he and his associates have in-
vestigated a variety of methods of measuring achievement motiva-
tion that do not involve imaginative stories. However, none of the
substitute methods proved to be successful, the criterion of success
being an appreciable correlation with achievement scores derived
from the content analysis of fantasy. Among the methods tried
were: {a) story alternatives to the pictures, presented in multiple-
choice form; (b) word stems to be completed with achievement
and nonachievement alternatives; and (c) agreement with achieve-
ment sentiments on a questionnaire. The only promising tech-
nique, one developed by Aronson (1956), is also a kind of projec-
tive or expressive test. This investigator finds that content analyses
of spontaneous doodles yield consistent significant correlations
with need-achievement scores over several samples of college stu-
dents. It is McClelland's (1956) belief that this method is poten-
tially valuable because it presumably can be used for children too
young to tell stories, for civilizations without written records, for
persons who are apparently too anxious about achievement to write
of it fluently, and for groups who speak entirely different languages.

In spite of these failures to find other useful techniques for meas-
uring the need for achievement, some progress has been reported
by investigators outside of the McClelland group, Elizabeth
French (1956), for example, reports some success in developing

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 251

a measure of complex motives that does not involve the use of
pictures. In her test, which is disguised as a test of insight, the
subjects are presented with ten single-sentence descriptions of the
behavior of hypothetical individuals. A sample sentence is "Ray
works much harder than other people," and the subjects are asked
to "explain" the behavior described therein. The rationale of the
test is that subjects who have, say, high achievement motivation,
will tend to project that motivation into their explanations of the
hypothetical instances of behavior. Some of the test items were
adapted from those used by Sheriffs (1948), and the method of
scoring the subjects' answers was similar to that used by McClel-
land. By altering the nature of the scoring categories, estimates of
either the achievement motive or the affiliation motive were said
to be obtainable. It is too early to know how this type of test will
compare with the more traditional method of measuring achieve-
ment motivation, but French has reported a significant correlation
of .48 between performance on a digit-letter substitution test and
test-defined achievement motivation under task-orienting instruc-
tional conditions.

After prolonged practice in the scoring of TAT stories following
the procedure of McClelland and his associates, a satisfactory de-
gree of interscorer agreement can apparently be obtained. Never-
theless, many workers have found that even after considerable
training scoring reliabilities remain low. To circumvent this dif-
ficulty an objective form of the TAT called the Iowa Picture
Interpretation Test (IPIT) was developed at the State University
of Iowa. In this test, TAT pictures are presented singly, each one
accompanied by a set of four verbal statements. Within each set
of statements, one is designed to reflect achievement motivation
and the other three to reflect insecurity, blandness, and hostihty.
The statements were selected as relevant to each of these four
categories on the basis of judgments by clinical psychologists. The
subjects' task is to look at each picture and indicate the degree to
which the accompanying statements are appropriate to the pic-
ture by arranging them in rank order. Scores are computed by
obtaining the sum of the ranks for each category.

Preliminary administrations of the test for standardization pur-
poses revealed that the test-retest reliability of the achievement

252 THE MOTIVATION OF BEHAVIOR

scores, though discouragingly low, was about the same as that of
the standard story form of the TAT when scored by skilled raters.
However, achievement imagery scores from the IPIT (Hurley,
1955) failed to correlate significantly (r = .12, N = 45) with
scores obtained by means of McClelland's system of scoring. Never-
theless, Williams (1955) has reported that subjects with high
achievement scores (IPIT-defined) worked significantly faster on
a simple test of addition than subjects with low scores. This find-
ing is similar to that reported by Lowell (1952b), using the
McClelland method of scoring, and is supported further by John-
ston's (1957) data showing that high achievement-imagery sub-
jects of both sexes attempted significantly more simple addition
problems over 10 two-minute trials than did subjects with low
achievement ratings. In addition, Johnston (1955) found that
high achievement subjects perform more efficiently on an elec-
trical maze task, and Hurley (1957) reports a positive relation
between IPIT achievement scores and the frequency of both cor-
rect and incorrect responses in a verbal learning task. It would
seem, therefore, that IPIT achievement ratings may reflect, in
part, individual differences in level of motivation. Nevertheless,
considerable research remains to be done before it can be con-
cluded that the IPIT is measuring a motivational variable rather
than something else or that extensive similarities exist between
whatever variable it does measure and other purported measures of
achievement motivation. Encouraging evidence for the view that
a good objective scoring method may eventually be developed has
recently been presented by Johnston (1957), who showed that the
test-retest reliability of the IPIT could be substantially improved
by lengthening the test.

Finally, it is worth noting that Hedlund (1953) has made a
serious effort to objectify the categories by which achievement
scores are obtained from subjects who are exposed to achievement-
type TAT pictures. He designed an objective multiple-choice
test, called the Iowa Multiple-choice Achievement Imagery Test,
which consisted of 70 three-foil items. One foil of the three was an
achievement-related statement and the other two were unrelated.
Ten such items were constructed for each of seven achievement-
motive-arousing pictures. The result of parallel experimental ad-

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 253

ministrations of both the objective test and of the standard story
form was that the internal consistency of both tests over pictures
was extremely low. That is, individual pictures did not tend in a
consistent manner to yield comparable estimates of the relative
need for achievement of different individuals. Moreover, neither
Hedlund's objective test nor the standard story form was related
in any clear-cut manner to several aspects of classroom perform-
ance or to performance on an anagrams test. In this latter respect
Hedlund's results failed to corroborate the relationship between
achievement scores and anagrams performance previously reported
by Clark and McClelland (1950). Hedlund concluded, therefore,
that neither the objective form nor the story form of the need-
achievement test had predictive utility for the specific performance
criteria he chose to investigate.

Motivational Differences Defined in Terms of Scores
on a Scale of Manifest Anxiety

In this section we shall consider theory and evidence relating
to the view that individual differences in level of general drive can
be usefully defined in terms of the responses of subjects to ques-
tionnaire items purporting to reflect manifest anxiety. This con-
ception, developed originally by Janet Taylor (1951), and by
Spence and Taylor (1951), has led to an extensive amount of
discussion and research, and only the barest outline of its many
details and ramifications can be presented here.

Taylor's method of defining human drive level by means of an
anxiety scale was grounded upon Hull's multiplicative-drive theory
and upon the results of studies of the acquired drive of fear in
animals (see Chapter 5). The extension of these principles and
relations to human behavior suggests that those who differ in
degree of fearfulness or anxiety should also differ correspondingly
in general level of drive. And if drive strengths differ, then be-
havior in diverse situations should be affected in a manner con-
sistent with the assumption that drive (D) is a nonspecific factor
affecting all reactive tendencies. Those who have studied the rela-
tion between questionnaire-defined anxiety and performance have
addressed themselves essentially to these problems.

254 THE MOTIVATION OF BEHAVIOR

Taylor's manifest anxiety scale (referred to hereafter simply as
the MAS) was constructed in the form of a questionnaire-type
personality inventory. Its component items, many of which were
adapted from the Minnesota Multiphasic Personality Inventory,
posed questions that clinical psychologists judged would elicit
answers reflecting manifest emotionality or anxiety. Few if any
of the individuals who have worked with the MAS, at least among
those in the Iowa group, have ever been concerned with the prob-
lem of whether the test measures "true anxiety," whatever that
may mean. Nevertheless, as Taylor (1956cz) has shown, MAS
scores correlate about as well with psychiatrists' judgments of
manifest anxiety and with other criteria as do other tests purport-
ing to measure anxiety. But these correlations are irrelevant to the
central aim of most of the studies involving the MAS. The mem-
bers of the Iowa group have simply assumed that it may be useful
to define drive level (D) in terms of MAS scores. They have then
proceeded to explore the utility of this definition by determining
whether performance in a variety of situations is affected in ways
that are consistent with the theoretical implications of this defini-
tion and of the system from which it springs.

Empirical Findings of MAS Studies. The first and still one of
the most striking findings in the MAS literature is that reported in
Taylor's original paper. She administered her test to members of
an undergraduate class and on the basis of their answers picked
out a high-drive (high anxiety) and a low-drive (low anxiety)
group. The high-drive group consisted of those who fell in the
upper 12 per cent of the distribution of MAS scores, and the low-
drive group was made up of those in the lower 9 per cent. When
both groups were run under identical conditions in an eyelid-con-
ditioning situation, it was found that the high-drive subjects con-
ditioned more rapidly than did the low-drive subjects. As the con-
ditioning curves (Fig. 7:3) for these subjects show, the anxious
subjects conditioned much more rapidly than did those in the
nonanxious group. Presumably the habit strengths of the two
groups were equal since both were given the same number of condi-
tioning trials. The difference between the performance levels of
the two groups is highly significant and is entirely consistent with
the hypothesis that the groups differed in general drive strength.

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE

255

Nonanxious Ss /
/

— ^

11-20 21-30 31-40 41-50 51-60 61-70 71-80
Trials

Fig. 7:3. Eyelid-conditioning curves for subjects with high (anxious) and low
(nonanxious) scores on a questionnaire designed to measure degree of mani-
fest anxiety. {From Taylor, 19 SI.)

Since the appearance of Taylor's paper, her principal finding that
eyelid conditioning is better with high-anxious subjects has been
confirmed in eight independent experiments (see Taylor, 1956d;
Spence, 1956; Spence and Ross, 1957). In all but one of these
studies the differences were statistically significant.

In a summary of experimental evidence bearing on the relation
of anxiety to performance level in eyelid conditioning, Spence and
Ross (1957) and Spence (1958) have pointed out that the agree-
ment among studies from the Iowa laboratory has been supris-
ingly good. Not only has the anxiety-level variable yielded rather
uniform results, but degree of conditioning has also been shown
to vary with the strength of the UCS, a variable which, as we
have noted above, is also presumed to affect drive level. Fig. 7:4,
reproduced from Spence (1958), illustrates these rather remark-
able interexperiment consistencies. The four points on the curve
labeled "random subjects" are from four independent experiments
in which randomly selected subjects were employed. As would be
expected from the theory, these points define a curve falling be-
tween the curves for high and for low anxiety. The data-points on

256

THE MOTrVATION OF BEHAVIOR

the high- and low-anxious curves corresponding to puff intensities
of 0.25 and 1.5 pounds per square inch are from an unpublished
experiment by Spence and Haggard, and the remaining four points
are from the Spence and Taylor (1951) investigation. Especially
impressive is the fact that although the three functions were ob-
tained by combining the results of six different experiments, the
functions all exhibit the same general negatively accelerated form.
The evidence from studies of classical defense conditioning,
then, is rather consistent in showing that high-drive subjects ac-
quire conditioned reactions more quickly than do low-drive sub-
jects. At least two experiments, however, one by Hilgard, Jones,
and Kaplan (1951) and one by Bitterman and Holtzman (1952)
have failed to support the expectations of the anxiety-drive theory
with respect to performance in classical conditioning situations.
In the former, although the high-anxious subjects gave more con-
ditioned eyelid responses than the low-anxious subjects, the dif-
ference was not statistically significant. And in the second study,
in which the galvanic skin response rather than the eyelid re-

5 40

■ 0 0.25 0.6 1.0 1.5 2.0

Intensity of unconditioned stimulus -Ibs/sq in.

Fig. 7:4. These data, assembled from a number of different experiments, in-
dicate primarily that the strength of conditioning increases with the intensity
of the unconditioned stimulus. Also to be seen are unusually consistent differ-
ences attributable to level of manifest anxiety. {From Spence, 1958.)

High

anxious Ss\

^^^.^^^

-^'

- ^

"^-Random Ss

^^ J i

/ 1

""^

^^,^^

/ f

^^"""^

/ /

' /

^ Low anxious Ss

/ F^

1 /

4 /

1 I

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 257

sponse was being conditioned, college subjects in the upper 50
per cent of the distribution of MAS scores were slightly but not
significantly superior with respect to conditionability. As Taylor
(1956cz) points out, however, the number of subjects in the first
experiment was unusually small, and in the second moderately
anxious subjects were included in both the high- and low-anxious
groups by the procedure of selection.

Anxiety, Chronic or Acute? Concerning the general concept of
anxiety and level of drive, it is of interest to note that two views of
the process have been proposed by members of the Iowa group.
Taylor, in her original article, suggested that subjects who scored
high on her scale might be considered chronically anxious. If so,
they would tend to behave as though they had a relatively high
level of drive even in nonstressful situations. Alternatively, Rosen-
baum (1953), and Spence, Farber, and Taylor (1954) have se-
riously entertained the possibility that persons who score high on
the MAS are individuals who, though not especially anxious when
performing simple, familiar tasks, become anxious or overreact
emotionally when tested in unusual or threatening situations. On
the first of these views, high anxiety is regarded as more or less
chronically present in certain individuals; on the second, it is an
acute emotional reaction or state aroused only in stressful kinds of
situations.

The experimental evidence bearing upon these two views of the
nature of anxiety is not yet sufficiently extensive to permit us to
choose between them, though the weight of evidence seems to
favor the acute theory. Rosenbaum, for example, in a study of
stimulus generalization used either a buzzer, a weak shock, or a
strong shock as a "punishing" stimulus for slow reactions. On the
acute hypothesis, anxious subjects should differ from nonanxious
subjects when a strong shock is used but not when a nonnoxious
buzzer is used, provided the situation is, in other respects, devoid
of threat. Rosenbaum found, in accord with this view, that the
differences between high- and low-anxious subjects were not sig-
nificant save under the strong-shock condition. Apparently the
weak shock was so mild — it was described to the subjects as a
"slight" skin stimulation — that it was no more stressful than the
buzzer. Further support for the acute theory comes from a study

258 THE MOTIVATION OF BEHAVIOR

of eyelid conditioning by Spence, Farber, and Taylor (1954). Anx-
ious subjects, who were conditioned by these investigators under
neutral conditions, showed only a slight (nonsignificant) supe-
riority in response frequency when compared with nonanxious
subjects. But when occasional shocks or threats of shocks were in-
troduced between trials, the anxious individuals performed at a
significantly higher level than those with low MAS scores. Finally,
the experimental findings of Bindra, Paterson, and Strzelecki
(1955) may also be interpreted as indirectly supporting the acute
hypothesis. In their study, involving classical salivary conditioning,
which is presumably nonstressful, the performance of the high-
and low-anxious subjects did not differ. The chronic theory would
predict, of course, that even in a nonemotional situation such as
this, the high-anxious subjects should, because of their higher
level of drive, perform more effectively, unless strong competing
(task-irrelevant) responses are dominant.

Studies such as those of Wenar (1954) and of Spence and
Farber (1953) can be interpreted as supporting either the acute
or the chronic view. In the first of these investigations, which was
concerned with the relation of anxiety to temporal generalization,
the experimenter measured the reaction times of his subjects to
a buzzer, a weak shock, and a strong shock. He failed to find,
however, any increase in the difference between the mean reaction
times of anxious and nonanxious subjects as a function of the in-
tensity of the stimulus to which they were reacting. The finding
that the groups differed even when reacting to the supposedly
neutral buzzer fits the chronic theory, and if the buzzer and the
two shocks are assumed to be equally stressful, the data could be
made to conform to the acute theory. The findings of Spence and
Farber (1953) compare with those of Wenar and require some-
what similar interpretations. These experimenters did not find the
difference between the conditioning scores of anxious and non-
anxious subjects to be a function of an increase in the intensity
of airpuff used as the unconditioned stimulus. Again, one might
try to support the acute theory by declaring that all puffs were
equally noxious, though this seems unlikely. But the fact that
even with the weakest puff the high-anxious subjects conditioned

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 259

more readily than those with low MAS scores is coordinate with
the chronic-anxiety theory.

Anxiety and Complex Learning Situations. Nearly all of the
MAS experiments we have discussed can be regarded as ones in
which few competitive tendencies to make task-irrelevant re-
sponses were present. In the eyelid-conditioning situation there is
doubtless some tendency for a subject to inhibit his blinking reac-
tions to the CS, and some writers (Hilgard and Marquis, 1940)
have maintained that conditioned responses always develop in
the face of an inhibitory set. Nevertheless, through the use of ap-
propriate instructions and by having the subject make a voluntary
blink to a ready signal, the typical tendency to hold the eye open
can be largely eliminated. When this has been done, it seems rea-
sonable to regard the situation as one in which the indicator re-
sponse is essentially the dominant member of the hierarchy. In
these cases, the theoretical expectation is that high-anxious sub-
jects should perform better than low-anxious subjects. But when
the task is such as to permit or support the appearance of strong
interfering responses, the theory becomes more complex and the
predictions change. It is to a consideration of these situations and
the theory appropriate to them that we now turn. Our treatment
of these matters rests in considerable degree upon that presented
by Spence (1956) in his Silliman lectures.

In a stimulus-response theory of the Hull-Spence variety, overt
performance on any task is a consequence of complex interactions
and competitions among task-relevant (correct) and task-irrelevant
(incorrect) response tendencies. Other factors equal, the stronger
the drive, the stronger all excitatory tendencies become, but also,
as a result of the multiplicative assumption, the greater the dispari-
ties among the absolute strengths of all members of the hierarchy.
If tendencies dominant under low drive are relevant to successful
performance of the assigned task, an increase in drive should lead
to improved performance. But when the dominant tendencies are
those regarded as incorrect by the experimenter, an increase in
drive will lead to the relative strengthening of these incorrect tend-
encies, and performance should become worse. This theory, as we
have seen, enables one to maintain that drive always functions as

260 THE MOTIVATION OF BEHAVIOR

an energizer, while providing an adequate explanation for the
observation that high drive, on occasion, leads to poorer rather
than to better overt performance.

In the earliest experiments designed to study the relation of
anxiety level to performance in complex situations, both verbal
(serial learning) and motor tasks (finger maze) were used. From
analyses of these tasks, it was predicted that high-drive (anxious)
subjects should perform more poorly than low-drive subjects. This
expectation was based on the known fact that, because of the
serial nature of these tasks, rather strong perseverative and antici-
patory tendencies are often present. Such tendencies — an intrusion
error in serial verbal learning is a good example — are, by definition,
task-irrelevant, and their presence should always tend to degrade
performance. Furthermore, the disruptive effect of these conflicting
tendencies should be greater for the high-drive than for the low-
drive subjects.

Within limits, these predictions have been borne out. Thus,
Taylor and Spence (1952), using a special type of serial-verbal
maze, and Farber and Spence (1953) with a finger maze, have
found that low-drive subjects tend to be the better performers. In
the second of these studies, however, it was noted that some of the
choice points elicited very few errors, and it seems reasonable,
therefore, to regard them as choices where the correct tendency
was dominant. At such points the high-anxious subjects should have
performed better than low-anxious ones, but they did not. Nega-
tive results have also been reported by Hughes, Sprague, and Ben-
dig (1954), who failed to obtain significant differences between
the performance levels of extreme MAS groups who were tested
with several serial-verbal mazes.

At this point appears the perennial problem of whether the
obtained effects of anxiety differences could be due in part to
factors other than differences in drive. Obviously they could,
and those who have worked with the MAS have been quite aware
that individuals lying at the extremes of the scale may differ not
only with respect to drive intensity but also with respect to a
variety of other traits or characteristics. One of the several pos-
sibilities that has been pointed out (Spence, 1956) is that people
who score at the high end of the MAS may be those who would

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 261

also rank high on a scale of "degree of susceptibility to distrac-
tion." Should this be the case, then anxious subjects may do more
poorly on serial learning tasks because they are paying attention
to stimuli that are irrelevant to successful task performance and
not because their drive level is high. This is clearly an associative
interpretation of performance differences, since it appeals to stable
individual differences in habits of "distractibility." Child and
Waterhouse (1953) have similarly sought to explain certain in-
stances of so-called motivated behavior by reference simply to
number and strength of task-interfering tendencies.

In order to be able to control and to manipulate the relative
associative strengths of task-relevant and task-irrelevant tendencies,
Spence, in his more recent studies (1956), has turned to the paired-
associates method of verbal learning. This task requires the subject
to form associations between the component members of each of
several pairs of words. Upon seeing the first member of each pair,
the subject must speak the second word (usually within about two
seconds) before it is presented. On successive runs through the
list of paired words, the order of the pairs is changed to minimize
the likelihood that associations from one pair to another will be
formed within the list. Of special importance is the fact that rather
wide variations in the strengths of the correct and incorrect tend-
encies can be experimentally produced. For instance, the strength
of the association between the stimulus and response elements of
each pair can be increased by picking out response words having
a high probability of being given as associations to the stimulus
words. Presumably, the stronger the tendency for the stimulus
item to call out its particular response word prior to training in the
laboratory, the less the likelihood that incorrect responses will be
dominant. Competitive responses can also be minimized by mak-
ing certain that the stimulus members of the different pairs are
neither synonymous nor formally similar.

Spence (1956) has performed a number of experiments designed
to explore the consequences of varying the strength of competitive
tendencies in a paired-associates task. Figure 7:5 shows the results
of one of these investigations in which competition between pairs
of words was minimized while the strength of the tendency for
each stimulus word to elicit its paired response word was maxi-

262

THE MOTIVATION OF BEHAVIOR

100

■50

High anxiety
iV=20

' p ^ Low anxiety

/ JV=20

6 7

Trials

10 11

Fig. 7:5. Performance of high- and low-anxious subjects on a verbal learning
task. The paired-associate items being learned were selected to minimize com-
petition between separate pairs of words and to maximize the tendency for
each stimulus word to elicit its paired response word. {From Spence, J 956.)

mized. As predicted from the theory, the high-anxious subjects
learned to perform this task significantly better than did low-
anxious subjects. Presumably this is because the lists were success-
fully structured to minimize the appearance of incorrect compet-
ing reactions, even though the task, taken as a whole, was relatively
complex.

Several additional studies have been reported by Spence, in
which increases in the strength of interfering tendencies were
deliberately produced by using response words that were unlikely
to be elicited by their paired stimulus words and by using stimulus
words that were synonyms for one another. With this type of
paired-associates list, the low-anxious subjects tended to perform
better on the strong-interference pairs than did the highly anxious
subjects. In this respect the results of three different experiments
were consistent, but in none was the effect statistically significant.

In concluding this section, we find it interesting to note that
Castaneda, McCandless, and Palermo (1956) have developed a
modified form of the MAS for use with fourth-, fifth-, and sixth-
grade children. Moreover, these investigators have shown that the
performance of anxious children on difficult tasks is significantly

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 263

worse than that of nonanxious children (Palermo, Castaneda, and
McCandless, 1956), that anxiety level is negatively correlated with
intelligence test scores, at least for girls (McCandless and Cas-
taneda, 1956), and that when task difficulty is varied there is a
significant interaction between anxiety and task difficulty, anxious
subjects doing relatively superior work on easy items and inferior
work on difficult ones (Castaneda, Palermo, and McCandless,
1956). Thus, the empirical relationships found for anxious and
nonanxious adults tend to be confirmed with children.

Summary

The present chapter deals with theoretical and experimental
aspects of the problem of how motivational variables affect the
performance of human subjects in a variety of situations. The ef-
fects on performance of drive, defined in terms of severity of
deprivation, in terms of exteroceptive stimulation and verbal in-
structions, and in terms of responses to standardized tests, are
considered in turn. To reduce the enormous literature in this area
to more manageable proportions, experiments dealing with the
effects of motivational variables upon perceptual behavior have
been treated separately in the following chapter.

Investigations in which attempts have been made to vary hu-
man motivation by manipulating severity of food deprivation are
relatively scarce, and the results are far from consistent. Accord-
ing to generalized drive theory, hunger should enhance all reactive
tendencies, thereby leading to better performance when correct
habits are strongest and to poorer performance when incorrect
tendencies are dominant. Experiments designed to evaluate these
expectations have seldom been conducted, however, and most in-
vestigators have proceeded on the assumption that hunger or other
needs should augment only those responses that are related to the
need. When subjects are maintained on a semistarvation schedule
for prolonged periods it is found that much of their waking time is
indeed devoted to thoughts and daydreams about food and eat-
ing. But efforts to obtain more objective evidence for this phe-
nomenon through the administration of standardized psychological
tests have generally been unsuccessful. Similarly, attempts to show

264 THE MOTIVATION OF BEHAVIOR

that relatively short periods of complete deprivation will enhance
the frequency of food-related responses to TAT-like pictures, to
word-completion test items, and to free-association stimulus words
have failed to provide consistent or conclusive support for the
hypotheses under investigation.

The administration of electric shock to human subjects either
before, during, or after the elicitation of certain responses produces
significant effects, which, in certain instances, are like those to be
expected of motivational variables. Specifically, reactions that are
consistently followed by shock reduction tend to be learned, and
those followed by shock induction tend to be abandoned. More-
over, randomly administered shocks sometimes appear to produce
an enhanced emotional or motivational level that is generally
facilitative of performance. Under a variety of conditions, however,
these generalizations do not hold. Not only does shock tend to
elicit responses that interfere with the successful performance of
some tasks, but it may also serve, particularly with human sub-
jects, as a source of specific knowledge concerning the correctness
or incorrectness of various responses. Although the motivational
consequences of other strong stimuli have not been extensively
studied with human subjects, we may note that the intense noises
associated with industrial or military operations do not, for the
most part, appear to function as motivational variables.

Attempts to manipulate human motivation by the use of "ego-
involving" or "success" and "failure" instructions have produced
diverse results. Where these instructions are relatively nonspecific
and are accompanied by improved performance, the arousal of a
general drive may be indicated. In other instances, however, in-
structions seem to function associatively, either because they con-
tain problem-solving hints or because they have a capacity to evoke
facilitating habits or attitudes. But few studies of this type have
been designed so that a choice might be made between motiva-
tional and associative interpretations.

The procedure of defining degree of motivation in terms of re.
sponses to standardized tests has been widely used with human
subjects, although the method has been criticized on the ground
that it is not truly experimental and that it is especially likely to
involve the confounding effects of other variables. Upon careful

MOTIVATIONAL VARIABLES AND HUMAN PERFORMANCE 265

analysis, however, these objections appear to be unsupportable. In
principle, adequately rigorous operational definitions of drive
can be based upon either manipulative or selective methods.

The assumption that human motive strength can be usefully
defined by analyzing the content of subjects' imaginative stories
underlies the extensive work of McClelland and his associates. For
the members of this group all motives are learned, and all are
affectively toned expectations of the coming of pleasant or un-
pleasant consequences. As determinants of behavior, motives ap-
parently always function as facilitators, although the effect is
limited to responses that are in some way relevant to the motive.
Moreover, the mechanisms involved in the facilitative action of
motives are simply those of associative learning theory, and no
explicit appeal is made to a concept of drive or motivation as such.
The theory is thus seen to be associative rather than motivational,
though the intricacies of the associative interactions have not been
presented in detail.

Another large group of investigations, in which drive level has
been defined by test-situation behavior, consists of the work of
Taylor and Spence and their collaborators. These investigators
have proceeded on the assumption that subjects with different
drive strengths can be selected on the basis of their responses to
a scale of manifest anxiety, and that the behavior of such subjects
in diverse test situations should be deducible from the basic rela-
tions of Hull's multiplicative-drive theory. Although failures to
confirm these expectations have been noted, the theory is well
supported by a substantial body of positive experimental findings.
Generally speaking, both adults and children who are defined as
having high drive perform more efficiently than do low-drive
subjects on tasks involving few competing responses. But where
strong interfering tendencies are present low-drive subjects tend
to be superior, precisely as predicted. The underlying theory may
be characterized as primarily motivational, since appeal is made to
a concept of nonspecific drive, but associative principles also play
an important supplementary role.

CHAPTER

iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiy^

Motivational Variables
and Perception

It has long been known that sensory and perceptual experiences
are largely determined by the properties of the stimulus energies
impinging upon sensory receptors. And every student of psy-
chology has some acquaintance with the specific stimulus-response
relations or psychophysical laws that have been painstakingly deter-
mined by psychologists and physiologists. Less firmly established,
however, and of more recent vintage, is the view that the reactions
of subjects in perceptual experiments depend not only upon stim-
ulus variables but also upon the personality, attitudes, associative
predispositions, and level of motivation of each and every subject.
In this chapter our attention is focused upon the last of these rela-
tions, that between motivational variables and behavior in situa-
tions described as perceptual.

The Problem of Perception in the Study of Behavior

When pressed for an exact account of the nature of perception,
psychologists rather generally tend to favor either of two somewhat

266

MOTIVATIONAL VARIABLES AND PERCEPTION 267

disparate views. The first of these views is that perception can be
construed most adequately as the private experience of the per-
ceiver. To understand what perception is Hke, it is only necessary
that one engage in perceiving for himself. Perception, according to
this phenomenological position, thus turns out to be whatever the
individual perceiver perceives perception to be. Those who favor
this interpretation appear to be enchanted by the wonders of per-
ception and regard the solution of its mysteries as the key to the
understanding of all psychological problems. On this view, all the
phenomena of learning, motivation, personality, adjustment, and
social behavior can be reduced to, or best understood as, examples
of the operation of perception.

The second, less sanguine view of perception begins by rejecting
the notion that perception can ever make scientific sense so long
as its essential nature lies hidden within the private experiential
world of the perceiver. Perceiving in someone else is not an activity
that is visible to the naked eye of the scientist. If one is to use the
word "perceiving" at all, therefore, with reference to organisms
other than one's self, the term must be introduced by means of a
formal definition, the elements of which are ultimately reducible
to directly indicatable things. One can observe the behavior but
not the experiences of one's subjects; hence any assertion about
the perceptions of these subjects must be based ultimately upon
how they behave. According to this second point of view, percep-
tion must have the status of a scientific inference or construct, and
like so many other constructs (e.g., motivation, habit strength, and
cognition), perception is real only to the degree that its presence
as a component of our explanatory accounts of behavior serves to
enrich our understanding of that behavior. Moreover, if percep-
tion is to function as an element in a systematic account of be-
havior it must be defined independently of the behavior it is pre-
sumed to be affecting, and it must have behavior-determining
properties, or relations to antecedents and other constructs, that
are different from those ascribed to other postulated entities or
processes. If not, then perception, which is inherently no less ex-
pendable than any other scientific term, need not be invoked as a
component of any behavior system.

Is a Concept of Perception Ever Needed? Those who hold to

268 THE MOTIVATION OF BEHAVIOR

the second of the above views tend also toward the opinion that
many, if not all, of the investigations commonly described as per-
ceptual can be carried to successful completion without appealing
to a concept of perception. This is especially true, it would seem,
of studies in which subjects respond immediately and overtly to
the presentation of stimulus objects of one kind or another. Sup-
pose, for instance, that subjects are asked to identify simple geo-
metrical forms. If certain responses are elicited by particular forms
in a regular manner, then empirical relations can be established
between physically definable stimulus objects, on the one hand,
and frequencies and/or kinds of responses, on the other. Subse-
quently, almost unlimited experimentation can be conducted to
determine the conditions under which these relations do or do not
hold. One can vary the size, color, shape, intensity, texture, and
distance of the figures, or note whether the responses are affected
by training, by the use of inverting lenses, by the level of dark
adaptation, and so on. The goal of such research is to determine
the range of conditions across which the stimulus-response func-
tion remains invariant; or, alternatively, to outline the boundary
conditions that must be fulfilled if the stimulus-response law is to
hold true. This can be done quite successfully without, at any
time, referring to or invoking a concept of perception. Innumer-
able studies could be cited to support the contention that a con-
struct of perception is not essential either to the formulation of
many stimulus-response laws or to the determination of the
interactive effects upon them of motivational, associative, and
other variables. Moreover, highly detailed and satisfactory physi-
ological interpretations of the manifold organic events interven-
ing between stimulus and response can be developed without in-
cluding perception as a link in the explanatory chain.

The preceding arguments suggest that the concept of perception
may be superfluous in many situations, but the possibility remains
that the concept might play a useful explanatory role in some
situations. One such situation is that in which a stimulus is pre-
sented but no overt response appears until some time after the
stimulus has been removed. In attempting to interpret this be-
havior one might assume that during the initial stimulus-exposure
period, a covert perceptual response or process was aroused. And

MOTIVATIONAL VARIABLES AND PERCEPTION 269

this internal, perseverating sequel to stimulation might then be
treated as the immediately effective determinant of the subse-
quently evoked overt response. This view has the virtue of permit-
ting us to incorporate into an objective theory of behavior any
instance in which a subject says that he perceived something (in
the past) even though at the time the stimulus was presented
he made no overt reaction whatever.

Defining Perception. If perception is to serve as a conceptual
bridge in such delayed-response experiments and in others, it must
be operationally defined, the definition being structured so as to
contain no reference to the particular behavior that is alleged to
be determined by perception. In defining intervening variables
such as drive, one can formulate acceptably independent defini-
tions in several ways, and the same variety of definition is ap-
parently acceptable for the concept of perception. To take a
single example, a subject might be defined as "perceiving" in one
situation on the basis of his behavior under other conditions. To
make this clear, let us suppose that we have determined for a given
subject an empirical relation between the verbal response "triangle"
and a physical stimulus having a triangular form. If this subject
is now presented with a triangle, but for one reason or another
makes no overt verbal response, we can still assert that he perceived
the triangle, by definition. These relations are shown in schematic
form in Fig. 8:1. The St Rr linkage at the top of this dia-
gram is the empirical law relating the triangular stimulus {St) to
the overt verbal response (Rf) • This law is then used as the basis for
the assertion that a covert perceptual response (rp) is evoked in the
stimulus-presentation situation where no overt reaction occurs. If,
as is shown in the bottom line of the diagram, the verbal response
is evoked subsequently when the subject is instructed to recall,
even though St is not present, then this response may be attributed
to the perceptual reaction. The occurrence of Rf is in no way es-
sential to the structure of the definition of Tp, however, and it can
be asserted, therefore, that the subject has perceived, irrespective
of whether Rf, some other response, or even no response at all is
elicited during the recall phase. Thus this definition of tp is seen
to be, as it must be, entirely independent of whatever behavior one
might choose to attribute to Tp.

270 THE MOTIVATION OF BEHAVIOR

Empirical law Sj, j >~Ry

Definitional chain

t
Stimulus present, Sj. ►- rp

no overt response

Recall situation-
overt response, Tp >-R^

stimulus absent

Fig. 8:1. This diagram summarizes the several steps and relations that would
be involved in one kind of definition of perception. First of all, an empirical
law relating the presentation of the triangle (St) to the occurrence of an
overt, identifying verbal response {Rv) must have been firmly established. This
law then provides the grounds for the statement that the perceptual response
(rp) has been evoked, by definition, when the triangle is presented but the
overt response is absent. Finally, it may prove useful to invoke Vp in interpret-
ing instances in which Rv appears in the absence of the triangle.

Inasmuch as perception can be defined in a variety of different
ways, the results obtained with one definition may differ sub-
stantially from those obtained with another. For instance, if a
child is presented with a number of circles and is told to draw
what he sees, he may draw a number of ovals. If these responses
are then used to define the child's perceptions, it is necessary to
conclude that circles are perceived as ovals. And this conclusion
must stand until some other definition of the child's ability to
perceive can be formulated, which can be shown to have more
scientific utility and significance than the first. For example, if the
child shows under other conditions that he is capable of sorting
circles into one pile and ovals into another, we may find it profita-
ble to redefine his perceptions accordingly. By one of these
criteria the subject perceives circles as ovals, but by the other he
does not. When two definitions of perception fail to coincide in
this manner, however, neither is "right" and the other "wrong."
A correctly formulated definition is "wrong" only in the sense that
in the interpretation of a broad range of behavior it may be less
useful than some other definition. The well-known psychophysical
methods, with their emphasis on the precise control and measure-
ment of stimuli, on repeated observations, and on the use of
statistically grounded inferences, provide us with our most trust-
worthy methods of deciding what is perceived and when.

MOTIVATIONAL VARIABLES AND PERCEPTION 271

Motivational Variables and Perception. Any experiment in
which subjects are asked to identify, detect, or compare physically
present stimuli would qualify, according to many writers, as being
concerned in some way with perception. Inasmuch as the subjects
in such experiments are responding overtly, it seems reasonable to
expect that the introduction of motivational variables might pro-
duce changes in the frequency, latency, and/or correctness of the
subjects' responses. The responses involved are the customary ones
of labeling, naming, and discriminating, by saying "same," "dif-
ferent," "larger than," and the like. Consequently, if one chooses y
to define perception in terms of these responses, one must conclude
that perception has been modified by a motivational variable when-
ever that variable is accompanied by changes in such responses.

Although experiments such as the above are often labeled "per-
ceptual," they possess no unique attributes or characteristics by
means of which they can be unambiguously distinguished from
the so-called "discrimination" or "choice" experiments. Moreover,
if perception is regarded as a covert response or process (e.g., r^ in
Fig. 8:1) intervening between the external stimulus (S) and an
overt response (R), and if a motivational variable is operating

throughout the entire S R sequence, then any effect of that

variable upon R may be attributable either to a change in the rela-
tion of S to Tp, or to a modification of the relation of rp to R.
Apparently, most studies of the effects of motivation on percep-
tion have been designed so that the motivational variable is present
both while the stimuli are being presented and while the response
is being evoked. None of these studies can be said, therefore, to
have answered the question of whether perception, as well as overt ^
behavior, can be modified by motivation.

In principle, however, an experiment designed to provide this
information appears to be feasible. For example, one might present
identical, impoverished stimuli to two groups of subjects, one
operating under high drive, the other under low. Overt responses
would be prevented by instructions during this initial stimulus-
presentation phase. Then at a later time the drive level of both
groups would be equated, and they would be given some kind of
recall or recognition test. If drive affects perception, the perform-
ance of the two groups should differ during the second test phase.

272 THE MOTIVATION OF BEHAVIOR

and such performance differences could not be ascribed to the
effects of drive upon the relation of r^ to R. Effects of this latter
variety could be detected, however, by equating drive during the
stimulus-exposure phase and varying it during recall.

From these considerations the conclusion is reached that
irrespective of how the motivation-perception problem is formu-
lated, perception cannot be said to have been affected by motiva-
tion unless behavior is affected. To study the role of motivational
variables in perception is to study the changes wrought by these
variables in the frequencies or kinds of responses exhibited by sub-
jects in so-called perceptual experiments. Whether these changes
in response are to be ascribed to modified perceptions or to changes
in response systems other than perception hinges on the design
and interpretation of one's experiments.

Multiplicative-drive Theory and Perceptual Behavior. Hull, in
theorizing about behavior in classical and instrumental condition-
ing situations, did not include the construct of perception in his
group of intervening variables. But the responses evoked in per-
ceptual experiments are clearly items of behavior, and it is ap-
propriate, therefore, to inquire as to how such responses might be
affected by changes in Hull's drive (D).
^ ^ Let us consider, by way of example, a psychophysical (percep-

tual) experiment in which a subject's absolute threshold for visual
stimuli is determined by the method of constant stimuli. With
this method, a number of discrete stimuli, whose intensities range
above and below threshold, are presented singly to a subject.
Usually he is instructed to say "yes" if he sees the stimulus and
"no" if he does not. The experimenter tabulates the frequencies
of "yes" and "no" responses corresponding to each of the stimuli
and after converting these frequencies into percentages plots them
against the stimulus values to obtain the usual psychophysical
function. The solid-line curve of Fig. 8:2 is a fictitious example of
such a function. As this curve shows, the hypothetical subjects
seldom perceive (say "yes" to) the faintest stimulus and almost
always say "yes" when the brightest stimulus is presented. By con-
vention, the absolute threshold is defined as the value of stimulus
intensity that elicits "yes" responses 50 per cent of the time. In
this graph, the threshold lies at stimulus 4 since the solid curve

MOTIVATIONAL VARIABLES AND PERCEPTION

273

100

75 -

Motivating--.

instructions ^^^

/ /
/ /
/ /

'C Usual
^instructions

A A
/A

/ Lncw

ttireshold
ginal threshold

1 1 1

<!- 25

12 3 4 5 6 7

Stimulus intensity

Fig. 8:2. The solid-line curve is a fictitious example of the kind of psycho-
physical function commonly obtained when the method of constant stimuli
is used to determine an absolute threshold. The dashed-line curve, with its
corresponding threshold, is the predicted outcome of the assumption that in-
structions alleged to be motivating actually serve to affect the associative
strengths of the "yes" response to all stimulus intensities while leaving dri\e
level unaltered.

intersects the horizontal 50 per cent response hne directly above
4. Responses of saying "no" are not plotted in the figure, but these
would decrease from a maximum at the left to a minimum at the
right and would form a mirror image of the "yes"-response curve.
To apply Hull's multiplicative-drive theory to observations such
as these we must first obtain some estimate of the relative habit
strengths of the responses being recorded. On the simplifying as-
sumption that drive does not vary with stimulus intensity, it
follows that the habit strength of the "yes" response must var}'
directly with the frequency with which that response appears. The
habit strength to say "yes" must be greatest, therefore, for stimulus
7 and weakest for stimulus 1. This fundamental assumption of a
direct relation between empirical response percentages and habit
strength is indicated by the "habit-strength" legend on the right-
hand ordinate. It is not necessary, for purposes of this analysis,
to specify the mechanisms by which these habits have been devel-
oped. It must be presumed, however, that these associative tend-
encies have been acquired prior to the experiment and are trans-

274 THE MOTIVATION OF BEHAVIOR

ferred to specific laboratory stimuli by means of verbal instructions
to the subjects.

Having made this assumption, we next turn to the question of

how this psychophysical function would be affected by the ma-

.nipulation of a motivational variable. Within the theory two

/ distinct answers can be given, depending on whether the variable

\ leads primarily to a change in habit strength or to a change in

^ drive. In this discussion we shall not consider instances in which

motivational variables might affect both associative strength and

drive.

To illustrate the first of these possibilities, suppose an attempt
is made to raise drive level by telling subjects to "try hard" and
to "pay very close attentio^n" to the stimuli and to the assigned
task. Instead of affecting drive, such instructions may function
simply to elicit specific receptor-adjustive acts (cf. Spence, l9S\a;
Wycoff, 1952) such as blinking less often, moving the head less,
and fixating the center of the exposure field more consistently.
Or they may evoke tendencies to repress thoughts of impending
social activities, classroom examinations, and so forth. If such
responses are elicited by the alleged motivating instructions, the
subjects should indeed show improved performance in detecting
faint stimuli. In this instance better performance would be indi-
cated by an increase in the frequency of /"yes" responses and a
corresponding decrease in "no" reactions. (And since,^by assump-
_tion, response frequency reflects habit strength, it must be con-
cluded that the instruction-induced adjustive acts have led to an
effective increase in the habit strength of the -l^yes" response for
all stimuli save those of maximum initial strength ./The observable
result of such an over-all increase in habit strength would be a
leftward shift of the psychophysical function and a corresponding
drop in the threshold. This is shown in Fig. 8:2 by means of the
dashed-line curve.

Several studies designed to test these expectations have been
conducted in the psychological laboratories of the State University
of Iowa. The results of two of these are shown in Fig. 8:3. The
data plotted in the upper panel were obtained by Bechtoldt and
McDonough (1958) in a situation in which the subjects were
asked to report whether a faint gray spot could be seen when

MOTIVATIONAL VARIABLES AND PERCEPTION 275

100

Motivated group

iV=15

Relaxed group

N = 13

-10

8-6-4-2 0 2 4 6 8
Exposure duration in ms. above and below threshold

lUU

^-0- ■"

(/>

Motivated attitude -

\ ' /

X^ /

' /

/ .^

/ ^^

w 50

tt)
>>

/^ 1

/ ,

/ /

^^Relaxed attitude

JV=20

_y^

1 1 1 1

-10 -8-6-4-2 0 2 4 6 8

Exposure duration in ms. above and below threshold

Fig. 8:3. The data plotted in the upper panel were obtained by Bechtoldt and
McDonough (1958) and those in the lower panel by Ludvigson (1958). In
both instances, attempts to increase the subjects' level of motivation by verbal
instructions shifted the psychophysical functions toward the left and lowered
the absolute thresholds. These empirical curves may be compared with the
theoretically predicted results shown in Fig. 8:2.

276 THE MOTIVATION OF BEHAVIOR

presented at very brief exposures in a tachistoscope. The right-
hand curve was obtained from subjects who were told to relax
and take it easy, and the left-hand curve from subjects who were
told to do their very best. Comparable results from an investiga-
tion by Ludvigson (1958) are shown in the lower panel. In his
experiment a single group of subjects performed a tachistoscopic
identification task under instructions designed to be motivating
at one time and nonmotivating at another. The results of both
of these studies are consistent with the hypothesis that "pay-atten-
tion" instructions can alter behavior in a perceptual experiment by
affecting associative tendencies. When instructions produce these
effects they should not, in all strictness, be described as "motivat-
ing," and interpretations of their influence on behavior should be
carefully identified as associative, since they contain no reference
to a motivational concept. Incidentally, there is little reason to
suppose that instructions to try harder will always lead to better
performance. Such instructions might actually arouse interfering
tendencies and thereby produce a decrease in efficiency.

The second answer to the question of how an alleged motiva-
tional variable might affect behavior in a psychophysical experi-
ment involves the assumption that the variable produces a change
in drive strength alone. Referring to the hypothetical data plotted
as the solid line in Fig. 8:4, it is apparent that the empirical
frequencies of "yes" and of "no" responses are exactly equal for
the threshold stimulus, and that therefore the associative strengths
must also be equal at that point. Under these conditions an in-
crease in drive should not produce any change in the relative fre-
^ quencies of the "yes" and "no" responses, since the corresponding
excitatory potentials will remain equal. Consequently, the threshold
in a psychophysical experiment of this kind should not be affected
if the motivational variable leads only to a change in drive.

When the "yes" and "no" habit strengths are not equal, how-
ever, an increase in drive, on the multiplicative assumption, will
magnify the absolute difference between the two excitatory tend-
encies and thereby increase the frequency with which the domi-
nant reaction is evoked. This condition of habit-strength inequality
is met for all stimuli save 4. At stimuli 5, 6, and 7 the "yes-
response" habit is stronger than the "no-response" habit, and an

MOTIVATIONAL VARIABLES AND PERCEPTION

277

100

Jfi 75

50 -

Q- 25

:3J^

High drive ---:v/ X

/ /
//
1/

Low drives

/ / '
// 1

^ / f
X/ /

// 1

^^ 1 /Threshold

1 2

3 4 5
Stimulus intensity

Fig. 8:4. If a motivational variable leads solely to an increase in drive, with-
out affecting associative strengths, then the accompanying theoretical analysis
suggests that the psychophysical function relating percentage of "yes" re-
sponses to stimulus intensity should become steeper. This is indicated here by
the dashed-line curve. The absolute threshold, however, should not be affected
by changes in drive alone.

increase in drive should enhance the frequency of "yes" responses
at those points. By the same reasoning, intensified drive should
reduce the frequency of "yes" responses to stimuli 1, 2, and 3,
since the dominant tendency at these points is to say "no." These
predicted effects, which are shown in Fig. 8:4 by means of the
dashed line curve labeled "high drive," should be negligible, of
course, at stimuli 1 and 7 because of "floor" and "ceiling" effects,
respectively. On the multiplicative-drive theory then, intensified
drive should increase the steepness of the slope of the psychophys-
ical function throughout the middle portions of the stimulus range,
but the absolute threshold, if defined in the conventional manner,
should not change. Since an increase in slope means a decrease in
res,ponse variability, this indicates, in one sense, an increase in "per-
ceptual sensitivity." This prediction may be less intuitively reason-
able than our first, and the writer knows of no experimental data
that clearly support it, but it does fit with the commonly held view
that lackadaisical (unmotivated?) subjects yield flatter psycho-
physical functions than those who attend strictly to business.
Under the conditions specified in the foregoing example, an

278 THE MOTIVATION OF BEHAVIOR

increase in drive per se should not reduce the absolute threshold.
But in the typical experiment on motivation and perception, the
threshold is not defined as the point where the stimulus is per-
ceived 50 per cent of the time, but as the point at which it is
perceived two or three times correctly in succession. This is equiva-
lent to defining the threshold as lying somewhere above the 50
per cent level, e.g., at 75 per cent. Referring to Fig. 8:4, we can see
that the intersection of the high-drive curve with the horizontal
line at the 75 per cent level falls to the left of the intersection of
the low-drive curve with the same line. If the threshold were thus
defined as lying above 50 per cent, then increased drive should
lead to a lower threshold.

The foregoing interpretations have dealt only with the effects
of motivational variables upon absolute thresholds, but similar
principles can be applied to experiments in which the method of
constant stimuli is used to determine difference limens. In such
experiments, each of several comparison stimuli is repeatedly paired
with a standard, the subject being asked to judge whether the
comparison stimuli are "greater" or "less" than the standard with
respect to some property such as loudness, length, brightness, or
weight. The introduction of a motivational variable into an experi-
ment of this kind ^ould be expected to produce an over-all dis-
_placement of the psychophysical function, provided the variable
leads to a general change in the strength of the associative tendency
to say "larger than" (or "smaller than"). The practical outcome
of such a displacement would be a shift in the point of subjective
equality and the appearance of a constant error. (In a subsequent
section of this chapter it is noted that the "value" of a stimulus
object might affect a psychophysical function in this manner.) The
slope of the function should not be altered, however, by habit-
strength modifications alone, save in the unlikely event that asso-
ciative tendencies corresponding to stimuli that are greater and
less than the standard are affected in opposite directions.

If the motivational variable enhances drive, leaving associative

jp -- strengths unchanged, then an increase in slope is the only pre-

^^^^^^^^ dieted outcome. Initially, associative tendencies to say "greater

than" should be stronger than tendencies to say "less than" for

stimuli greater than the standard, and the reverse should be true

MOTIVATIONAL VARIABLES AND PERCEPTION 279

for stimuli less than the standard. An increase in driye. should
function, therefore, on the multiphcative assumption, to increase
the excitatory potentials of correct responses relative to incorrect
ones, thereby reducing the size of the difference limens and im-
proving accuracy of discrimination. Conversely, decreased drive
should result in less accurate discriminations. However, the point ,
of subjective equality, where "greater than" and "less than" re- v,
spouses are equally frequent and their associative strengths are
presumably equal, should not be affected by modifications in drive
strength.

Tlie results of several studies conducted at the State Universit}
of Iowa are consistent in showing that the psychophysical function
tends to become steeper under high drive when the method of
constant stimuH is used to determine difference limens. The data
from two of these experiments, by Clark (1958) and by Tandler
(1958), are shown in Fig. 8:5. The same task, that of judging
whether each of a number of comparison circles was larger or
smaller than a standard, was used in both studies. And in both
experiments drive level was supposedly lowered by pretending that
the subjects' responses were not being recorded on certain trials
and that these trials served only as rest-period activity. It is clear
from these experiments that under the alleged high-drive condition
the frequency of larger-than responses for stimuli larger than the
standard was increased and the frequency of larger-than responses
for stimuli smaller than the standard was diminished. However,
the issue of whether the observed effects are due to changes in
drive per se or to modified habit strengths is not settled by these
results.

Throughout the foregoing theoretical development, it was as-
sumed that the stimuli were relatively simple events such as faint
lights or tones. In most experiments relating perception to motiva-
tion, however, words, nonsense syllables, or complex forms are
used instead of lights or tones, and the subjects are required to
name or identify the stimuli. When such stimuli are presented
clearly and for relatively long periods of time, the correct responses,
with rare exceptions, are evoked 100 per cent of the time. Hence
if one is to study the effects of motivational or other variables on
perception, the stimuli must be presented unclearly or in some

280

THE MOTIVATION OF BEHAVIOR

-5 -

-3-2-1 0 1 2 3
Difference between variable and
standard stimulus, mm

100

Normal drive

^•f

iV=32

Reduced / /
drive ^ / /

J /

^^ /

--r^^

1 1 1 1 1

1 1 1

B. F. Tandler

-5-4-3-2-1 0 1 2 3 4 5

Difference between variable and

standard stimulus, mm

Fig. 8:5. This figure presents the results of two independent but identical ex-
periments (Clark, 1958; Tandler, 1958) on the role of motivation in visual
judgments of area. The method of constant stimuli was employed to
determine difference limens, some observations being made under "normal"
conditions and others (reduced drive) when the subjects were led to believe
that their responses were not being recorded.

MOTIVATIONAL VARIABLES AND PERCEPTION 281

impoverished form so that identification is not always perfect.
Impoverishment can be effected by blurring the outlines of visual
forms, by shortening exposure times, by reducing stimulus in-
tensity, or by various combinations of these methods.

To extend the multiplicative-drive theory to responses elicited
by such impoverished stimuli, it is necessary first to consider the
effect of impoverishment upon the habit strength of the identifying
response. One possible assumption is that impoverishment pro-
duces no change other than a decline in habit strength. If so, then
increased drive should magnify the values of the excitatory poten-
tials for all impoverished forms of the stimulus, and performance
in identifying the stimulus should improve. Moreover, this con-
clusion should hold for any stimulus to which a naming response
has been strongly associated, irrespective of whether the stimulus
is or is not related to the drive-producing conditions.

Alternatively, impoverishment might lead not only to a decline
in the strength of the habit for the correct response, but also to a
rise in the strength of other competitive habits. For instance, if
the printed word house were made fainter and fainter, the middle
letter might disappear sooner than the rest. Hence, for the im-
poverished stimulus, the strength of the habit to say "hose" might
become stronger than that to say "house." Should impoverish-
ment have this effect, accuracy of identification should decline
under high drive, since the associative tendencies to utter the
wrong words would be stronger than those to speak the correct
words. The effect of drive upon the absolute threshold for simple
physical stimuli (see Fig. 8:4) may also be regarded as an instance
of habit-strength reversal of this kind, the habit strength for a
response of "no" being presumed to exceed the "yes" response
habit for all subthreshold stimulus values.

These expectations from the multiplicative-drive theory may or
may not prove accurate when subjected to experimental test. For
our present purposes, however, this is of minor importance. Of
more significance is the fact that these limited predictions illustrate
the application of a motivational theory to behavior which, if one
chooses, may be described as "perceptual." While these guesses
have been made without reference to a concept of perception, it is
entirely possible that performance in psychophysical tasks will
ultimately demand such reference.

282 THE MOTIVATION OF BEHAVIOR

Facilitative effects of motivational variables upon identifying re-
sponses to impoverished stimuli can be interpreted associatively as
well as motivationally (cf. Brown, 1953d). Through past learning,
need-related responses are more likely than others to have become
associated with stimulus aggregates composed of both internal and
external cues. When the strong supportive function of unambigu-
ous external stimuli is weakened through impoverishment, the
relative importance of internal cues will be enhanced, and if the
motivational variable alters these internal stimuli, selective facili-
tative effects may be observed. In ever)' instance, therefore, where
motivational variables appear to have modified the responses that
define perception, it is imperative to inquire whether the effect
was indeed motivational or whether it could have been due to
existing differences among, or changes in, the relative strengths of
associative tendencies.

Primary Sources of Drive and Perception

The studies evaluated in this section are necessarily limited to
those involving hunger and thirst as primary sources of drive for
the reason that almost no other primary sources of motivation
have been manipulated in studies of perception. Electric shock,
which also seems to qualify as a primary source of drive, has often
been used in perceptual experiments but in a manner that makes
the treatment of such studies more appropriate to a later section
on perception and secondary sources of motivation. Specifically,
electric shocks have nearly always been paired with visual stimuli
during preperception phases of experiments, and have usually not
been administered while perceptual proficiency is being measured.

The Effects of Deprivation upon Perceptual Responses. An ex-
periment by McClelland and Atkinson (1948) provides us with
an example of investigations of the effects of food deprivation
upon behavior in a perceptual situation. In this study sub-
jects who had gone without eating for periods of from 1 to 2,
4 to 5, and 16 to 18 hours were asked to report what they saw at
various times on a dimly illuminated screen. On most occasions
no pictures of any kind were projected on the screen, but the
subjects were made to believe that extremely faint pictures were

MOTIVATIONAL VARIABLES AND PERCEPTION 283

indeed being shown. This behef was strengthened by the initial
showing of a recognizable object and by occasional hints from the
experimenter as to the nature of the "objects" being shown. Under
these conditions, subjects who had been without food for 16 to 18
hours gave significantly more food-related responses than did those
with 1 to 2 hours of deprivation, the number of responses made by
the 4 to 5-hour group falling in between. The results were not
entirely consistent, however, since in some cases, in answer to
questions posed by the experimenter, increasingly hungry subjects
gave more frequent responses related to instrumental acts of secur-
ing food, but did not give more frequent responses dealing with
food items as such. Moreover, as the experimenter's hints contained
increasingly direct references to food, the number of food responses
increased, but not differentially as a function of hunger. And when
vaguely defined blots were projected on the screen instead of
blank slides, the frequency of food-related responses diminished.

According to some critics, the McClelland-Atkinson investiga-
tion is not "perceptual," since when blank slides were projected on
the screen there was "nothing there" save a dim "blob" of light
for the subjects to perceive. Nevertheless, the subjects were told
that very faint pictures would be projected and that guesses might
often have to be made as to the nature of the pictures. Under
these conditions the subjects did report seeing things, and, with
the exceptions noted above, the more severe their hunger the
more numerous their food-related responses. Thus an empirical
relationship was obtained between a motivational variable and re-
sponses by which perception could be defined. The question of
whether it would be scientifically useful to define perception in
terms of responses occurring in the absence of structured external
stimuli can be answered only by a careful evaluation of the relative
merits of this and of alternative definitions.

Lazarus, Yousem, and Arenberg (1953) have reported that the
recognition of tachistoscopically presented pictures of food objects
improves with hunger. This conclusion was based on data obtained
from four groups of subjects whose tachistoscopic recognition
thresholds for both food and nonfood objects were tested under
0 to 1, 2, 3 to 4, and 5 to 6 hours of food deprivation, respectively.
The trend of the data for the first three groups was clearly in a

284 THE MOTIVATION OF BEHAVIOR

direction to support the authors' conclusions, but a dechne in
performance was observed in the case of the hungriest group.
Because of this reversal, conclusions from this study as to the
sensitizing effect of hunger on perception must be limited to short
periods of deprivation. Nevertheless, the tendency for recognition
accuracy to improve with mild degrees of hunger was observed
in two independently conducted experiments, and somewhat sim-
ilar findings have been reported by Levine, Chein, and Murphy
(1942). According to Lazarus et al. the drop in perceptual effi-
ciency at the longer deprivation intervals may be a consequence
of a conditioned hunger cycle, a possibility previously suggested
by Sanford (1937).

That the facilitative effect of hunger on perception is relatively
ephemeral is suggested by the results of a further study by Lazarus
et al. This investigation was identical with the first, save that fewer
subjects were run in each group and all observers were required,
after each stimulus presentation, to select a response from a list
containing the names of the 10 pictures plus six additional items.
When freedom of choice was restricted in this way, recognition
of food objects became increasingly worse, though not significantly
so, as the period of deprivation was lengthened. No evidence was
found in either study to indicate that hungry subjects make more
food-related, prerecognition responses or give more food associa-
tions to the stimulus words of a free-association test than non-
hungry subjects. Thus the findings of these several studies, though
suggestive of a relation between hunger and tachistoscopic recog-
nition thresholds, are not, because of reversals of direction and
interexperiment inconsistencies, entirely convincing.

Two additional experiments, by Wispe and Drambarean and
by Taylor have yielded contradictory results. Wispe and Dramba-
rean (1953) asked subjects who had been without food and water
for 0, 10, and 24 hours to identify tachistoscopically presented
words. Both common and uncommon words, some related and
some unrelated to food and water deprivation, were used. Analyses
of recognition thresholds revealed that degree of "commonness"
was the most important factor determining word recognition.
Among both the common and uncommon words, however, signifi-
cant interactions were obtained between severity of deprivation

MOTIVATIONAL VARIABLES AND PERCEPTION 285

and need relatedness of the stimulus words. That is, recognition
times for neutral words were unaffected by deprivation, whereas
times for need-related words dropped markedly from the 0-hour
to the 10-hour condition, thereafter remaining at about the same
level for the 24-hour group. The authors conclude from these data
that need-related words are recognized more easily as deprivation
increases. Although the subjects' previous experience with words
would account for their ability to recognize common words more,
readily than uncommon ones, the apparently positive effects of
deprivation could scarcely be ascribed to the factor of familiarity
since the differently deprived groups were presented with the same
words.

Taylor (1956b), however, in an experiment similar to that of
Wispe and Drambarean failed to confirm their findings. In her
study the subjects' set to expect the presentation of food words
was manipulated in addition to their degree of physiological need.
In two replications of the experiment, with different orders of
stimulus presentation, Taylor observed that the recognition thresh-
olds of deprived subjects for need-related words did not differ
significantly from those of satiated subjects. Subjects who had been
set to expect need-related words, however, did have lower thresholds
than control subjects.

A series of experiments by Gilchrist and Nesberg (1952) pro-
vides some final examples of studies relating need to perception.
These investigators attempted to avoid certain of the methodolog-
ical deficiencies of previous experiments by studying the short-term
perceptual recall of both need-related and neutral objects. Specif-
ically, their subjects were shown colored slides of food objects (or
neutral objects), the intensity of the projector lamp being set at
a standard level. Subsequent to the projection of each slide, the
lamp voltage was either raised or lowered and subjects were asked
to readjust the voltage until the projected image looked the same
as when first presented. A stable result from an integrated series
of four experiments was that all subjects tended to make positive
time errors. That is, they tended to make the projected image
brighter than the standard level. The positive time errors for
satiated subjects remained constant, whereas those for deprived
subjects increased progressively with hours of food deprivation up

286

THE MOTIVATION OF BEHAVIOR

to a limit of 20 hours. In effect, therefore, performance became
progressively worse as a function of increasing hunger. In one
experiment involving pictures of fluids, increasingly severe thirst
(for subjects who were also hungry) was accompanied by higher
and higher voltage settings, whereas for satiated subjects no such
progressive increase in positive time error appeared. Figure 8:6
shows the lamp-voltage settings made by thirsty and satiated sub-
jects as a function of time since the start of the experiment.

Of special interest here is the dashed-line portion of the curve
for the thirsty subjects. As the label indicates, the final point on
the curve was obtained after the thirsty subjects had been per-
mitted to drink as much ice water, orange juice, and/or milk as
they wished. This strikingly precipitous drop was highly significant
and was replicated in two further experiments. In these, Gilchrist
and Nesberg showed that the increasingly positive time error
associated with increased need was not due merely to the visually
patterned nature of the food pictures but was dependent specifically
on their need-related characteristics. From these studies the authors
concluded that as need increased, positive time errors in the
illuminance matches of objects relevant to that need tended to
increase.

Since Gilchrist and Nesberg offer no theoretical interpretations
of their rather dramatic findings, it is of interest to see whether

0 2 4 6

Hours from start of experiment

Fig. 8:6. Average lamp-
voltage settings made by
thirsty and satiated sub-
jects in adjusting the
brightness of a projected
image of thirst-related ob-
jects to subjective equality
with a previously seen
image. True physical
equality would have been
obtained if the voltages
had been set at 80. The
drop at the right-hand side
for the thirsty group
coincided with their hav-
ing drunk fluids ad libi-
tum. (From Gilchrist and
Nesberg, J 952.)

MOTIVATIONAL VARIABLES AND PERCEPTION 287

the type of explanation proposed earlier in this chapter can be
expanded to encompass their results. We note first of all that
even satiated subjects consistently made positive constant errors in
their illuminance matches. This means that if two equally bright
stimulus objects are presented, one after the other, the second is
judged dimmer than the first. Or it means that if a subject is
allowed to adjust the brightness of a second object to subjective
equahty with a first, he tends to set the second brightness level
higher than that of the first. Thus a positive time error implies
that the tendency to make adjustive response settings above the
"true" value is stronger than the tendency to make adjustive reac-
tions below the "true" value. Hence the two corresponding habit
strengths must be unequal. But, as we have repeatedly noted,
when habit strengths are unequal, increasing drive produces an
increasing disparity between the corresponding excitatory tenden-
cies and thereby an increase in the predominance of one overt
response over another. Thus the finding that positive errors increase
in magnitude with hunger but not under satiation follows as a
logical consequence of the multiplicative-drive hypothesis. On this
view, drive variations should have no effect when constant errors
are absent. This theory does not, of course, encompass the data at
the 0-hour point in Fig. 8:6. However, the surprisingly low voltage
settings made at that point by the subjects who were later to
become thirsty may well have been due to sampling fluctuations.
At the start of the experiment both groups were equal with respect
to level of deprivation, and hence their lamp-voltage settings
should have been approximately identical.

This interpretation, though applicable to the difference between
deprived and satiated subjects, must be modified to fit the data
obtained by Gilchrist and Nesberg when the illuminance levels
of need-related and non-need-related pictures were matched by
the same subjects. These findings, shown in Fig. 8:7, would prob-
ably call for the introduction of an incentive motivation factor.
Thus, as deprivation becomes more severe, the need-related items
may elicit anticipatory responses of enhanced vigor, and greater
and greater increments might be added to drive. This should lead,
as indicated above, to an increase in the relative strength of the
dominant response and hence to progressively larger constant

288

THE MOTIVATION OF BEHAVIOR

Hours from start of experiment

Fig. 8:7. Additional data from the Gilchrist-Nesberg (1952) experiment. In
this phase a single group of subjects was tested with both need-related and
non-need-related stimulus objects. Quite clearly, the magnitude of the positive
error increases for need-related items but not for neutral items as a function of
hours of water (and food) deprivation. The effect seems to disappear im-
mediately, however, when the subjects are permitted to drink.

errors. By contrast, non-need-related items should not produce
incentive-generated motivational increments. At this point, how-
ever, the theory seems to fall short; for although increasing thirst
should have progressively magnified the constant errors even for
non-need-related items, it did not do so (see Fig. 8:7, lower curve) .
Nevertheless, the terminal drop of the curve for non-need-related
pictures, like that for need-related stimuli, is consistent with
theoretical expectations.

We conclude this section with the following summarizing com-
ments. Research purporting to have shown a relationship between
intensity of physiological need and responses in perceptual situa-
tions has yielded suggestive and stimulating, though controversial
and inconsistent, results. In some instances, deprivation seems to
facilitate the identification of need-related pictures or words pre-
sented at near-threshold levels. Rather often, however, inconsistent
or negative results have followed from the use of similar methods.
Indeed, in experiments such as those of Gilchrist and Nesberg,
increased deprivation has been shown to lead to poorer rather than

MOTIVATIONAL VARIABLES AND PERCEPTION 289

better "perceptual" performance. All in all, therefore, experi-
mental findings in this area, as well as theories as to their nature
and relationships to other varieties of motivated behavior, must
be viewed with reservations.

Secondary Sources of Drive and Perception

Up to this point, we have dealt only with experiments in which
deprivation has served as the principal motivational variable, and
in which motivational properties of the to-be-perceived stimuli,
if any, have tended to play a secondary role. By far the largest
group of experiments, however, consists of those in which the
stimuli serve both as to-be-perceived cues and as the principal de-
terminants of level of motivation. In nearly all of these cases the
presumed capacity of the cues to function as motivational variables
can reasonably be attributed to processes of learning. Such studies
can legitimately be treated, therefore, as instances of the influence
of secondary or learned sources of drive on perceptual behavior.

Motivation and Judgments of Magnitude. Most investigators
whose studies fall in this category have been concerned with the
problem of whether visual estimates of size are significantly affected
by the value of the objects being judged. Usually, the stimulus
objects have been coins, subjects being asked to equate coin size
to that of a neutral disk, whose diameter can be varied. Unfor-
tunately, many of the early experiments were characterized by
methodological, experimental, and statistical inelegancies, and
little confidence can be placed in the results. For example, in
some studies an illuminated patch of light served as the variable
stimulus, the diameter of which was adjusted by the subject to
equal that of a relatively dark coin seen against a light background.
This deviation from conventional psychophysical practice was
sometimes further magnified by permitting subjects to hold the
coins in their hands, tactual cues thereby being added to one stimu-
lus complex but not to the other. Moreover, adequate controls
were seldom included for the effects of the inscriptions on the
coins or for differences among them in color, thickness, and weight.
In spite of these defects, or perhaps because of them, the results
of early studies (e.g., Bruner and Goodman, 1947) were so strik-

290 THE MOTIVATION OF BEHAVIOR

ing that, if they did nothing else, they aroused widespread interest
and stimulated others to perform parallel experiments. In subse-
quent investigations the introduction of methodological refine-
ments has, in some instances, reduced or obliterated the supposed
enhancing effect of value on size. But positive results have also
been obtained under more carefully controlled conditions and the
phenomenon may well be genuine.

Although the majority of experiments have been directed toward
determining the effects of value on estimates of size, it has also
been reported that judgments of weight, numerosity, and bright-
ness are altered by so-called motivational factors. The motiva-
tional state is most commonly assumed to be aroused by the
specific stimuli being judged, though it has sometimes been varied
by the selection of subjects whose socioeconomic needs are pre-
sumed to differ.

Value as a Source of Drive. One of the major conceptual prob-
lems attending studies of the perception of valued objects is that
of specifying the mechanism by which variations in value might
produce changes in drive. Inquiries as to the nature of value raise
many complex questions that have intrigued and baffled economists
and philosophers for generations. For the student of behavior,
however, it seems apparent that value cannot meaningfully be
treated as a property of physical objects as such, and that the most
useful definition would be structured in terms of an object's power
to elicit preferential approach and seizing responses from organ-
isms. This would imply, incidentally, that value is always relative.
A lump of sugar has more positive value to a rat than a block of
wood if and only if the rat approaches and seizes the sugar instead
of the wood. A 50-cent coin is more valuable than a dime only
to the degree that individuals prefer the former to the latter.
If the individuals are infants rather than adults, or if they have
not been raised in our culture, or if the test is conducted during
a period of severe inflation, neither coin may have more value
than the other, and neither coin may have more value than a
blank disk of paper or metal. To say, then, that a coin or any
other object has value is to assert that it elicits responses of certain
kinds, at certain times, and from certain individuals, and in addi-
tion, that these responses are different from those elicited by other

MOTIVATIONAL VARIABLES AND PERCEPTION 291

coins or other objects. A physical object is valuable only to the
extent that organisms approach and choose it in preference to
other objects.

Defining value in terms of behavior helps to clarify the problem
of how a valuable stimulus object might produce an increase in
drive. The behavior of approaching and seizing objects such as
coins, jewelry, new cars, and the like, which is characteristic of
adults in our culture, is evidently learned as a consequence of the
rewards provided by society. When a subject in an experiment
on judgment of size is presented with both a coin and a neutral
disk, the coin must, if it is to be defined as having positive value,
elicit stronger tendencies of approaching and choosing than does
the neutral disk. And since the coin has become a positive goal
object, the sight of it may well elicit learned tendencies to make
anticipatory goal reactions (rgS), which in turn may contribute
increments to the subject's general level of drive (D). When the
problem is structured in this manner, value becomes covariant
with strength of incentive motivation, a topic discussed in detail
in Chapter 5.

In brief, when the value of a stimulus object is manipulated,
simultaneous (and correlated) changes are presumed to be induced
in the strengths of learned tendencies to approach and possess
the object. And in so far as anticipatory goal reactions occur,
their vigor, and hence the drive increments they provide, should
also increase with the value of the goal object.

A parallel interpretation can be made of studies in which
motivational or affective levels have been manipulated by the
selection of groups alleged to differ in personality or in social
position. Thus, if judgments of coin size are made by poor and
rich children, it might be argued that the groups differ with respect
to the strengths of their habits to seek and secure coins, and per-
haps also, therefore, in the intensities of their anticipatory [drive-
contributing) responses. However, variations in drive strength alone
should not affect judgments of coin size if the associative tendency
to say "larger than" is equal in strength to the tendency to say
"smaller than" when a coin is compared to a neutral disk of
identical size. Presumably, therefore, differences in the strengths
of linguistic habits must be invoked as an adjunct to the preceding

292 THE MOTIVATION OF BEHAVIOR

interpretation. This matter will be treated in more detail after the
consideration of additional illustrative investigations.

Experimental Studies of the Judged Size of Coins. If the value
of an object determines its apparent size to any degree, the effect
must be revealed through changes in the nature of subjects' re-
jponses concerning the size of that object. Phrased in this way,
the problem seems relatively straightforward, but the design of an
experiment adequate to provide unequivocal results is far from
simple. It would not do, for example, to present a coin to a subject
and ask him simply to estimate its size in inches or millimeters.
Even if the mean of his judgments differed significantly from true
physical size, one would have no way of knowing whether the
discrepancy was due to value, to the presence of a picture on the
coin, to the subject's lack of practice in making such estimates, or
to any of a number of other factors. In principle, there is only
one way in which the effects of value on judged size can be
unambiguously assessed: judgments must be made of the relative
size of a valued object when it is compared with less valued
objects having essentially identical physical properties. By this
criterion, an ideal experiment on size as a function of monetary
value, would be one in which subjects are required to say whether
genuine coins (or bills) looked larger or smaller than counterfeit
moneys. Were an experiment done in this way, all sources of
variability save value would be properly controlled. Conclusions
based on the use of control objects whose physical properties or
backgrounds deviate substantially from those of the valued objects
cannot be regarded as definitive. Moreover, for complete confirma-
tion of the value-size hypothesis, increases in value would have to
be accompanied by progressively greater distortions of apparent
size, at least over a certain range of sizes. The greater an object's
value, the greater should be the distortion of its apparent size when
compared with suitable control objects. Unfortunately, nearly all
existing studies have failed, in one respect or another, to meet
these ideal conditions, and we must content ourselves with sug-
gestive rather than conclusive findings.

Confining our attention for the moment to studies of common
coins, we find scant evidence that when they are physically present
their size is enhanced. In their initial investigation Bruner and

MOTIVATIONAL VARIABLES AND PERCEPTION

293

Goodman (1947) reported astonishingly large overestimations of
coin sizes when compared with judgments of cardboard disks. But
findings of subsequent studies in which the controlling base line
has been more appropriately provided by metal disks have, in the
main, been negative.

Carter and Schooler (1949), seeking to check the Bruner-Good-
man results, asked children to adjust the size of a circular patch
of light to correspond to the sizes of various coins, metal disks,
and cardboard disks. Their results, for coins and aluminum disks
only, are presented in Fig. 8:8. The cardboard-disk data have been
omitted on the ground that cardboard disks are less suitable as
control objects than aluminum disks. As these curves reveal, the
sizes of the penny, dime, and nickel were all underestimated,
though not significantly, both with reference to true size and to
aluminum disks. Both the 25- and 50-cent pieces were overesti-
mated, the latter significantly so. Subsequently, Bruner and Ro-
drigues (1953) also found the penny and nickel to be judged
smaller than worthless metal counterparts, only the quarter being
estimated as larger (nonsignificantly) than its corresponding metal
disk. Both coins and metal disks, however, were judged as larger

Monetary value

3 10-

E
2 5

5-5

^-10

10^ 1^ 5(^

25^

50^

18 20 22 24 26

Diameter, mm

28 30

Fig. 8:8. Experimental data indicating that small coins tend to be judged
smaller and large coins larger than aluminum disks equal in size to the various
coins. These results were derived from tests in which physically present coins
and disks were matched to a circular patch of light, whose diameter could be
controlled by the subjects. {Adapted from Carter and Schooler, J 949.)

294 THE MOTIVATION OF BEHAVIOR

than paper disks. The rather commonly held view that coins,
when physically present, tend to be judged larger than metal disks
of comparable size is thus apparently supported by the data for
only one coin (50-cent piece) from a single experiment (Carter-
Schooler). And even this effect may be due less to value per se
than to other factors such as the design on the coin.

Whether available data support the notion that degree of over-
estimation of size increases with value is also questionable. In the
Carter-Schooler study, the effect was indeed greater for the more
valuable of the two coins that were overestimated. But in the
Bruner-Rodriguez study, since only the quarter was overestimated
( the 50-cent piece was not used ) relative degree of overestimation
cannot be determined. These latter authors, finding little evidence
for absolute overestimation, suggest that when the members of a
stimulus series are valuable, the differential enhancement between
the terminal members of that series will be greater than between
corresponding members of a neutral series. But their data, cited to
support this contention, do not show differential enhancement,
if this is taken literally to mean that all coins are enhanced in
some degree, higher valued ones being enhanced more. Rather,
their findings, like those of Carter and Schooler, are that negative
^-^ I enhancement (underestimation) occurs for small coins and posi-
tive enhancement for larger ones. To explain this by the value
hypothesis one would have to contend that coins of little value
have even less value than metal disks; but such a twist would im-
pose a severe strain on the value hypothesis. Apparently, therefore,
it is necessary to conclude that when judgments are made of the
size of physically present coins, under controlled conditions, their
value is an as-yet-unproved determinant of their apparent size.

Before leaving this problem of judged size of coins, we should
note that the socioeconomic status of the subject, his age, and
other factors may influence his estimates. Children described as
"poor" by Bruner and Goodman overestimated physically present
coins more than did "rich" children. But Carter and Schooler,
who used larger groups, obtained no significant differences between
the judgments of rich and of poor children under coins-present
conditions. To confound the issue further, Rosenthal (1951) re-
ports the size estimations of rich ten-year-olds to be greater than

MOTIVATIONAL VARIABLES AND PERCEPTION

295

those of poor subjects for every coin, but the reverse to hold for
six-year-olds. In many studies, judgments have also been made of
the remembered sizes of coins by both rich and poor subjects.
These memory judgments have not been treated here since they
fall outside the conventionally defined area of perception, but it
appears likely that value has a more marked effect upon judgments
of remembered size than upon the apparent size of physically
present objects (e.g.. Carter and Schooler, 1949).

Other Studies of Apparent Magnitude and Value. We turn now
to other studies, in which, for the most part, coins were not used
as stimulus objects or in which unusual techniques were adopted.
In this latter respect the investigation of Ashley, Harper, and
Runyon (1951) is of special interest. Seeking to control attitudes
towards money, these investigators told hypnotized subjects to
forget their previous life histories. On some occasions the subjects
were told they really had lived very poverty-stricken lives, and at
other times that they had been reared in extremely well-to-do
homes. All subjects were also tested in the normal waking state.
When the "poor" state was induced, subjects tended slightly to
overestimate the sizes of all coins, relative to true size, and when
the subjects were "rich," they tended to underestimate all sizes.

The results of this experiment, for responses made with coins
present, are shown in Fig. 8:9. The differences between the sub-
jects' responses when "poor" and when "rich" were statistically
significant, but the differences between "poor" and "normal,"

Fig. 8:9. Judgments of the
size of physically present
coins made by subjects un-
der normal conditions and
under hypnotically in-
duced "rich" and "poor"
attitudes. {Adapted from
Ashley, Harper, and Run-
yon, 1951.)

i.l
1.0

Poor^y/^^

Normal^/ /

0.9

A /\X >

/^\ / / /

/ \ / / /

0.8

~ y /K \ / / ^

' / \ \ / / /

/-^V^x/^

0.7

\ /

0.6

A = True size

r , ,

5 10

Value of coin, cents

296 THE MOTIVATION OF BEHAVIOR

"poor" and "true" size, "rich" and "normal," and "rich" and
"true" size were not evaluated by Ashley et al. and were pre-
sumably not significant. Indeed, for three of the coins, the subjects,
when "poor," came closer to the true sizes than when "normal,"
and the size exaggerations produced by the "poor" instructions,
though they occurred for every coin, were of doubtful significance.
Moreover, the tendency of the subjects (adults) when in the
"rich" state to underestimate all sizes is consistent with neither
Rosenthal's data from ten-year-old subjects nor with the results
provided by Bruner and Goodman's rich subjects. Very possibly,
as Ashley et al. have noted, their findings may be due in part to
the fact that their subjects when "poor" were very attentive and
hence more accurate, whereas the same subjects when in the
"rich" state were relatively careless, responded quickly, and only
"condescendingly cooperated."

A final phase of the Ashley, Harper, and Runyon study marks
the use of a new and ingenious procedure. In this phase the
hypnotized subjects, under both "rich" and "poor" instructions,
were required to match the diameters of a variable circle of light
and a metal slug. At diverse times the slug was described to the
subjects as either lead, silver, white gold, or platinum. The results
of this test, reproduced in Fig. 8:10, show that estimated size
tended to increase with the declared value of the metal and that
"poor" judgments inclined more toward overestimation than "rich"
ones. Because the same slug was used under all conditions, one
cannot appeal to uncontrolled external stimulus factors to explain
the results. Unfortunately, since we have so little knowledge of
what takes place during hypnosis, these findings cannot readily
be generalized to "normal" subjects. Nevertheless, tests conducted
under hypnosis may reveal the operation of processes whose effects
in waking subjects are usually obscured by sets, attitudes, and the
like.

Experimental results entirely negative with respect to the size-
value hypothesis have been reported by Lysak and Gilchrist ( 1955 ) .
Using paper currency rather than coins, they required adult subjects
to match the bills against control rectangles bearing designs of
varying complexity. Estimates of size were found to be unaffected

MOTIVATIONAL VARIABLES AND PERCEPTION

297

Lead

Silver White gold

Alleged metal of slug

Platinum

Fig. 8:10. Subjects under hypnotically induced "rich" and "poor" attitudinal
states made judgments of the size of a gray iron slug. The effects upon their
judgments of being told that the slug was made of lead, silver, white gold, or
platinum are shown in the figure. {Adapted from Ashley, Harper, and Runyon,
1951.)

by the monetary value of the bills, though apparent size increased
with increasing complexity of patterns on the control rectangles.

In some experiments essentially valueless objects have been
artificially endowed with symbolic "value" by inscribing them with
marks such as the dollar sign. This procedure was followed by
Solley and Lee (1955). In their study white cardboard disks bear-
ing the dollar sign were found to be significantly overestimated
even though the nonmonetary symbols used on control disks had
been judged to have the same degree of "perceptual closure" as
the dollar sign. These data were interpreted to mean that symbolic
value affects perceived size, but the brevity of the original report,
which describes only one other symbol (swastika), makes critical
appraisal difficult. The disks were equated for degree of closure,
but this may not have eliminated the kinds of stimulus variables
which, in perceptual illusions, lead to nonveridical report.

The hypothesis that symbolic value affects apparent size is
supported to some degree, however, by a study of Dukes and
^van (1952). Their subjects participated in a "gambling" game

298 THE MOTIVATION OF BEHAVIOR

involving cards of identical size but of different symbolic value,
as indicated by printed numbers ranging from — 300 through 0 to
+ 300. These cards were randomly chosen, and the subject won
or lost amounts determined by the printed numbers. After each
such trial, the subjects selected, from a graduated series of blank
control cards, one which they judged equal in size to the value card.
As the monetary value printed on the test card increased from 0
to + 300 the subjects tended to select larger reference cards. That
is, the estimates of size were smallest for test cards of 0 (printed)
value and increased in a negatively accelerated manner as symbolic
value was heightened. However, the subjects also tended to select
larger and larger blank reference cards as the value of the printed
cards changed from 0 to — 300. If one regards this as decreasing
positive value, as seems reasonable, then the results for the negative
cards fail to conform to the value-size hypothesis. These results
might be due in part to the fact that the physical length of the
numbers stamped on the cards increased with both positive and
negative value. Consequently, the tendency to overestimate the
size of the cards at the extremes of the scale, whether positive or
negative, may have been dependent on the size of the printed
numbers rather than on "subjective value." No nonvalued control
cards were used to evaluate the significance of this factor.

Additional positive results have been reported by Beams (1954),
who observed that children with strong food preferences tended
to make significantly more frequent larger-than judgments of their
favored foods. Mausner and Siegel (1950), however, using stamps
of similar size and shape, but of alleged values ranging from 5
cents to $12, found no evidence for the alteration of recognition
thresholds by value.

Similar studies, in which factors other than value have been
manipulated, have produced both positive and negative results.
Thus Bruner and Postman's (1948) subjects judged tokens in-
scribed with positive symbols as larger than tokens bearing negative
or unpleasant inscriptions. Klein, Schlesinger, and Meister (1951),
however, in a comparable study were unable to demonstrate that
the apparent size of neutral objects was changed by affective
symbols. Young children may be more susceptible to these effects,
since Lambert, Solomon, and Watson (1949) report enhanced

MOTIVATIONAL VARIABLES AND PERCEPTION 299

estimates of the size of neutral disks that have been associated
with rewards, and diminutions of size when rewards are with-
drawn.

Though lack of space does not permit a full report of other
studies relevant to the problem of size as a function of value, it
seems fair to conclude that, at best, the effect is relatively slight
especially when the valued objects are physically present. Value
seems to exert its greatest effects upon the judgments of children
and hypnotized adults, and individual differences in the need for
money influence judged size, provided that estimates are made
from memory.

An Interpretation of Size-Value Relations. The data we have
reviewed on the size-value hypothesis, though leaving many ques-
tions unanswered, are sufficiently suggestive to warrant further
research as well as the development of theories as to how value
might affect apparent size. In this latter area little has been done,
and Jenkin (1957), in his comprehensive review, lists only two
such attempts. One of these is a single sentence by Bruner and
Rodrigues suggesting that the effect may be due somehow to the
frequent pairing in the environment of size and value. The other,
a proposal of Dukes and Bevan, is that value, like certain param-
eters of electronic circuits, functions to "tune" the organism to
respond selectively and with "amplification" to valued objects. In
an earlier section of this chapter it was suggested that valued
objects might function after the manner of incentives to arouse
anticipatory drive-producing reactions. However, the details of
the process through which value might affect size-estimating be-
havior were not presented, and it is to this task that we now turn.

In a psychophysical study employing the method of constant
stimuli, accentuated size would be revealed as a displacement of
the function toward the region of the stimulus dimension com-
posed of large comparison objects. Such a shift has been dia-
gramed in Fig. 8:11, the assumption being that the solid-line curve
was obtained when a metal disk the size of a 50-cent piece was
compared, as the standard stimulus, with other metal disks. The
dashed-line curve is assumed to have resulted from the use of a
real half dollar as the standard stimulus. In this imaginary experi-
ment, the point of subjective equality (PSE) falls at stimulus 4

300

THE MOTIVATION OF BEHAVIOR

(O T3
(O (D

Neutral disk as— -.^^ y/^ y^

standard stimulus ^^^V*^ ^

• /

/ ^

/ jt-- *^°'" ^^

/ /'^—-standard

/ / stimulus

/ \ /^

/ / 1

/ 1 / 1

X ^ '

/ ./jPSEforl
/ X 1 disk |-«—PSE for coin

^^^ ^

^^^ ^ 1 1

— ^T — <?• 1 1 1 1 1

1 2 3 4 5 6 7

Diameter of variable stimulus disks

Fig. 8:11. Fictitious data to illustrate the kind of displacement of a psycho-
physical function that should be obtained with the method of constant
stimuli if a coin tends to be judged larger than a neutral disk of equal diameter.
It is assumed here that the point of subjective equality {PSE) shifts from
variable stimulus 4 to stimulus 5 when the coin is used as the standard
stimulus.

with a metal-disk standard stimulus and at stimulus 5 with the
coin as a standard. For simplicity, no change in slope is assumed
to have accompanied the rightward displacement of the curve. In
interpreting the dashed-line curve we must remember that the
decreased percentages of "larger than" judgments for all the
comparison metal-disk stimuli mean a general increase in the
tendency to judge the valued coin as "larger."

As a result of our previous analyses the conclusion was reached

that an increase in drive alone should increase the steepness of a

psychophysical function, but should not displace the entire func-

rtion in the manner shown in Fig. 8:11. Altered habit strengths,

V however, could have this effect. The necessary steps in developing

y this argument are as follows:

First, we reiterate the view that an object such as a coin has
positive value only because individuals approach and select it in
preference to other objects. Consequently, objects can only be
rank-ordered with respect to value when we know the rank orders
of their capacities to elicit choosing and seizing responses. Sec-
ondly, in a great many instances, objective size tends to vary direct-

■ Pt

MOTIVATIONAL VARIABLES AND PERCEPTION 301

ly with value, especially when similar types of objects are grouped
together. Houses, cars, diamonds, ranches, TV sets, candy bars,
and bundles of money are, for the most part, more valuable when
large than when small. Consequently, it seems reasonable to expect
that tendencies to make verbal responses such as "large," "impres-
sive," "bigger," "heavier," "broader," and "higher," should become
strongly associated through learning with those objects toward
which strong habits of approach are being established. Hence it
follows that any object of demonstrable value must be one tending
strongly to evoke instrumental choice reactions and, in addition,
verbal responses of the general class "large." Finally, a shift of the
point of subjective equality must occur if a valued object elicits
more "larger-than" responses than a neutral object of equal size.
To find such a shift is to obtain empirical evidence that "value l^-J^H"^
leads to a perceptual accentuation of size."

It will be seen that this interpretation of the effects of value
upon judgments of size is purely associative, since it contains no
reference to a specific motivational entity. Moreover, this view is
similar in some respects to the Bruner-Rodrigues conception, men-
tioned earlier, and to suggestions presented by Lambert, Solomon,
and Watson (1949). Our interpretation is also consistent with
the findings of experiments such as that of Proshansky and Murphy
(1942), in which perceptual judgments were altered by rewards
and punishments.

The Perception of Stimuli Associated -with Noxious Events. As
we have seen, it may be possible, by manipulating the value of
a stimulus object, to vary habit strengths and/or incentive motiva-
tion and hence to change the nature of the responses defining
perceived magnitude. If a subject is repeatedly punished rather
than rewarded for responding to certain stimuli, one would expect
that avoidance behavior and emotionality would become strongly
associated with such stimuli and that changes in drive and habits
resulting from such learning might also affect perceptual behavior.

Existing experiments in which anxiety-arousing stimuli have been
used as the to-be-perceived cues may be divided into two broad
classes. In one the perceptual identification of initially neutral
cues is tested following a training period in which noxious stimuli
— usually electric shocks — have been paired with the neutral cues

302 THE MOTIVATION OF BEHAVIOR

— usually words. Studies constituting the second group are distin-
guished by the use of stimuli such as vulgar words, which, as a
consequence of socialization training, have supposedly acquired
tendencies to evoke emotional reactions of disgust, apprehension,
or anxiety. (The present section deals with only the first group,
the second being covered subsequently.) Though superficially dif-
ferent, both groups of experiments pose the question of whether a
conditioned stimulus for an emotional response can be more {or
less) readily identified than an equally familiar neutral stimulus.
Or, when account is also taken of typical experimental procedures,
the question becomes that of whether responses of identifying and
naming impoverished forms of a given stimulus will be altered
when that stimulus also elicits an emotional response.

Conceptualizing the problem in this manner permits us to relate
this kind of perceptual research to other traditionally nonpercep-
tual problems. For one thing, it becomes clear that in this area
the student of perception and the student of conditioned fear or
anxiety are dealing with much the same phenomena, though the
former individual is less interested in the strength of the emotional
response than in its effects on identifying reactions. The student
of perception asks whether a learned emotional tendency functions
as a motivational variable in modifying the effective strength of a
nonemotional associative tendency (the identifying response) when
both tendencies are attached to the same stimulus. But this, as
has been noted in Chapter 5, is very similar to the acquired-drive
problem. It is this relation that has led us to treat these perceptual
tasks as activities involving secondary sources of drive.

Studies of the perception of threatening stimulus objects are
also similar to experiments on stimulus generalization, save that
in the former studies the stimulus in its clearly identifiable form
elicits both a learned identifying response and an emotional reac-
tion that may function motivationally. In presenting generalized
(impoverished) versions of that stimulus, therefore, one seeks to
determine whether the generalized identifying response will be
modified in motivational ways by the simultaneously elicited gen-
eralized emotional response. A so-called neutral control stimulus is
presumably one whose impoverished versions elicit a generalized
identifying response but little or no emotion.

MOTIVATIONAL VARIABLES AND PERCEPTION 303

The problems that arise in studying the identification of inimical
cues are also related to those that attend research on stimulus
distinctiveness. The central problem in the latter investigations is
whether the identifiability of a stimulus, as estimated from the
ease with which new responses can be attached to it, can be
enhanced by associating other distinctive (often nonverbal) re-
sponses with that cue. Here, however, in contrast to the so-called
perceptual experiments, the added associative connections are
typically neutral emotionally.

In the light of the preceding analysis, and assuming that emo-
tional arousal often has motivational-variable-like effects, we must
go on to specify the mechanism or mechanisms underlying emo-
tionally altered perceptions.

Mechanisms of Emotionally Modified Perceptual Behavior. It is
an unfortunate fact that in the studies under consideration one
finds few specific statements or hypotheses as to how emotionality
may affect perception. Although it is usually assumed that emo-
tionality is motivating, almost nothing is said about either the
nature of motivation or its relation to perception. In general terms,
motivation is alleged to function either to prevent the individual
from seeing the stimulus clearly — this is termed perceptual defense
— or to help him to see the stimulus more clearly — this is percep-
tual vigilance or sensitization. As numerous writers have noted,
however, specific hypotheses as to which of the two effects will
occur under a given set of conditions are singularly rare.

On the assumption that a perceptual situation involves the
sequence of the stimulus (S), a covert perceptual response (fp),
and an overt (usually verbal) identifying response (Ry), there
appear to be two possible ways in which Rr can be affected by
stimulus-evoked emotionality. First, if the emotional response (re)
occurs either before, or concomitantly with, r^, then the develop-
ment of Tp, and ultimately of Ry, could either be hindered or
facilitated by r^, depending upon the properties assigned to fg.
Second, if Tg has a longer latency than rp, as would certainly be
the case if the occurrence of re were contingent upon rp, then any
effects of re upon overt behavior would have to be due to modifica-
tions of the functional linkage between rp and Rv or to direct
effects upon Ry. According to some authors, it is only when

304 THE MOTIVATION OF BEHAVIOR

behavior is affected in the first of these ways that one can properly
speak of "pure" perceptual defense or vigilance. For such writers
alterations in behavior attributable to the second of the above
mechanisms should be regarded as instances of response suppres-
sion or augmentation, since the alleged perceptual response (r^) is
not directly affected by rg.

To clarify the implications of this interpretation, let us suppose
that a single taboo word is presented just once at near threshold
level to each of a group of subjects. If this word is identified less
often than an equally familiar neutral word (presented in the
same way to other subjects), either the hypothesis of "pure" per-
ceptual defense or the hypothesis of response suppression could
account for the results. But if each subject is presented with only
one word on a single occasion, the emotionality produced by that
word can only affect r^, and thereby produce "pure" perceptual
defense or vigilance, if Tg occurs as soon as or before r^. Unfor-
tunately, little can be said about the relative latencies of emotional
and perceptual responses. But from what is known concerning
the latencies of peripheral autonomic reactions (cf. Solomon and
Wynne, 1954) it would appear that emotional responses may
develop too slowly to affect perception itself in a single-trial situa-
tion such as this.

Stimulus- induced emotionality may not, therefore, affect percep-
tion in this kind of single-trial experiment, but the procedures
followed in actual experiments are such that even if Vg does have
a long latency, it nevertheless can affect perception. Thus it is
customary in studies of perceptual defense to present each word
a number of times, exposure durations being increased progres-
sively, as in the ascending order of the method of limits, until
recognition occurs. Obviously even a long-latency Ve, provided it
persists for several seconds, can, under these conditions, antedate
and thereby modify perceptual responses elicited by subsequent
presentations of the same or different words. Moreover, the emo-
tionality generated by additional taboo words in a list could serve
either to heighten the average emotional level or to prevent its
decline. Conventional experimental procedures, therefore, insure
that Tp can be affected by emotionality, though they do not insure

MOTIVATIONAL VARIABLES AND PERCEPTION 305

that the influence of fg will be confined to Vp alone, since a persist-
ing Te could also affect the relation of tp to Ry.

Evidence to support this hypothesis of persisting emotionality
is provided by the McGinnies and Sherman (1952) study, in which
recognition thresholds for neutral words were found to be elevated
by the prior presentation of taboo words. They used the term
"generalization" for this effect, but it seems more appropriate to
describe it as persisting or residual emotionality. Incidentally, the
hypothesis of persisting emotionality leads to the methodological
recommendation that neutral words should not be interspersed
among nonneutral words in the same list if one wishes to maximize
differences in the thresholds for the two kinds of words.

Turning now to the specific roles that might be played by Te,
it is clear that both its possible contribution to general drive and
its associative effects merit consideration. On the assumption that
drive varies directly with the strength of rg, the problem is reduced
to the more general one that we have already discussed of word-
identification thresholds as a function of drive intensity. Thus if
the associative strengths of the neutral and taboo words in a single
list are simply weakened by impoverishment rather than sup-
planted by other (incorrect) associative tendencies, then all thresh-
olds should be lowered by heightened drive. Or if taboo words
alone are presented to one group of subjects and neutral words
to another group (familiarity equated) then only the taboo-word
thresholds should be lowered, provided additional qualifying as-
sumptions such as the above are also met.

Any associative interpretations that one might formulate of the
effects of Te upon word-identification thresholds would probably
follow the general lines of reasoning presented in Chapter 4 and
in earlier portions of this chapter. For example, it might be postu-
lated that Te leads to, or produces, distinctive internal stimuli (Se)
and that either facilitative or competitive reaction tendencies have
been, or can become, attached to Sg through learning. From these
assumptions, defenselike or vigilancelike effects, that are either
selective or nonselective with respect to taboo and neutral words,
could be satisfactorily explained. It should be noted, however,
that the concept of emotionality is not crucial to associative inter-
pretations such as these, and in a later section of this chapter it

306 THE MOTIVATION OF BEHAVIOR

will be shown that some of the findings of perceptual experiments
can be deduced from nonemotional assumptions resembling those
of modern conflict theory.

The Perception of Stimuli That Have Been Paired with Shock.
We turn now to specific laboratory studies in which recognition
thresholds have been determined for stimuli with which electric
shocks have been associated. Investigations of the recall of shocked
words or syllables are omitted on the ground that they are related
only indirectly to perceptual behavior.

In an experiment whose major purpose was not the investigation
of perceptual accuracy as a function of conditioned emotionality,
Lazarus and McCleary (1951) nevertheless obtained some relevant
evidence. In the standardization phase of their experiment, 10 five-
letter nonsense syllables were presented tachistoscopically at near
threshold levels. On the basis of the subjects' responses, two groups
of five syllables each were equated with respect to the number
of times they were correctly recognized and also with respect to
the frequency of their use by the subjects. Thus possible differences
in familiarity between the two sets of syllables were minimized.
Then, with the syllables clearly exposed for one-second periods,
galvanic skin responses (GSRs) were conditioned to five of the
syllables by the use of electric shock. Finally, all syllables were
again presented in a tachistoscope at short exposure times, GSRs
were recorded, and five seconds after each exposure the subjects
made verbal reports of what they had seen.

The results obtained from tabulations of verbal reports are of
immediate relevance here. These data, when corrected for response
preference, failed to indicate significant differences in the recog-
nition of shock and nonshock syllables. However, when no correc-
tion for response frequency was made, seven of nine subjects iden-
tified the shock syllables more accurately than the nonshock
syllables, though again the differences were not significant. These
data are consistent with the view that success in perceptual iden-
tification depends, in part, upon response frequency, but they
yield no support for the supposition that the addition of an emo-
tional tendency to a stimulus (by conditioning) modifies the
accuracy of identifying responses.

From analyses of their GSR data, Lazarus and McCleary con-

MOTIVATIONAL VARIABLES AND PERCEPTION 307

eluded further that even when subjects are unable to make correct
verbal identifying responses, autonomic discrimination may be
possible. This judgment was based on the fact that all nine of
their subjects gave larger GSRs to shock than to nonshock syllables
even when the syllables were incorrectly reported verbally. This
effect was termed subception to indicate that "perceptual" dis^
crimination, defined by autonomic activity, apparently takes placp,
even when "perceptual" discrimination, defined by verbal report,
does not.

The subception effect and comparable effects reported by Mc-
Ginnies (1949) and others have generated considerable argument
and counterargument. The findings place some theorists in the
difficult position of having to hold that a shock syllable is in a
sense "perceived" as a shock syllable (and hence the subject reacts
to it emotionally), and yet that it is not "perceived" since the
subject cannot identify it correctly. Thus subjects are both discrim-
inating and not discriminating.

This seeming paradox disappears if we reject the notion that all
behavior, including autonomic activity, is a reflection of perception
and that the "real" nature of perception can be appraised only
through phenomenological statements concerning conscious aware-
ness. In perception, as in all other subject-matter areas of psychol-
ogy, there is no one "true" measure, and inferences based on any
particular measure need not, and often do not, correlate perfectly
with inferences grounded in a second measure. A subject in a
verbal learning experiment, for example, may appear to have
learned nothing if we estimate his learning by means of the method
of recall. But with the savings method, we may be able to show
quite clearly that practice has led to considerable learning. Simi-
larly, a rat in a T maze may choose one side as often as the other,
and from this we would conclude that no differential response
tendency has been established. Measurements of running time,
however, may indicate significantly longer times on one side than
on the other, and quite different conclusions may be indicated.
By analogy, the subjects of the Lazarus-McCleary experiment were
not "discriminating" when verbal reports were taken as the criteria
of differential response, but when reference was made to the
average amplitudes of their autonomic reactions, they were "dis-

308 THE MOTIVATION OF BEHAVIOR

criminating." The fact that the two measures yield different results
should cause little surprise in the light of our experience in other
research areas. Bewilderment ensues only when we speak in non-
operational terms of "awareness." Unless we are to abandon the
principle of scientific determinism we must maintain, as Bricker
and Chapanis (1953) have insisted, that when responses to shock
syllables are genuinely greater than those to nonshock syllables,
and if other factors are equal, then the stimuli provided by the
two kinds of syllables must be unequal. The fact that verbal-identi-
fying responses to shock and nonshock syllables do not differ is
indeed interesting, but it is paradoxical only when we champion
the view that such responses constitute the only "true" mirror of
perception as it "really" is.

Returning now to additional experiments involving electric
shock, we find several of interest. Murphy (1953), for example,
obtained evidence that shock may facilitate the perceptual iden-
tification of nonsense syllables. However, the tachistoscopic thresh-
olds for shock syllables were not significantly lower than those
for nonshock syllables, save when a normal scale-value trans-
formation was made of the original data. Thus the degree of
generality of his results may be somewhat limited. Lysak (1954)
has also reported that previously shocked nonsense syllables are
more readily seen than neutral syllables if punishment is not ad-
ministered in the recognition situation. When shocks are given
both during training and recognition phases, however, significantly
higher thresholds are obtained for shock syllables than for non-
shock syllables.

Recce (1954) has described an investigation purporting to
demonstrate a relation between the reinforcement occasioned by
shock reduction following the pronunciation of a nonsense syllable
and the visual recognition threshold for that syllable. He used a
paired-associates verbal learning task in which electric shocks were
administered on half of the pairs of syllables. For one group of
subjects shock was terminated (shock-escape condition) as soon
as the response syllable was pronounced. But for a second group
the shock was kept on (nonescape condition) for the entire dura-
tion of syllable presentation. On the tachistoscopic tests the
shock-escape group exhibited generally lower thresholds than the

MOTIVATIONAL VARIABLES AND PERCEPTION 309

group for whom shock escape was not possible. The thresholds for
the shock-escape group, however, did not differ from those of a
nonshocked control group. Apparently the thresholds of the shock-
escape group were not lowered by reinforcement, but rather the
thresholds of the nonescapable-shock group were raised. During the
course of the paired-associates learning task, shock escape produced
significantly better learning than did the inescapable shock con-
dition, but not better than the nonshock condition. Thus the
obtained differences in thresholds appeared to be due, not to
improvement resulting from shock escape, but to interference
generated by inescapable shock. Poorer performance during learn-
ing would indicate weaker associative strengths, and hence the
recognition scores for the nonescapable-shock group should be
lower. (This would qualify, evidently, as an associative, not a
motivational, interpretation.) Finally, Recce found no significant
differences between the recognition thresholds for shocked and
neutral syllables within each group, regardless of whether the
syllables were the "stimulus" or the "response" members of the
pair. This is consistent with the negative findings of the Lazarus-\
McCleary study described above, and with the theoretical analysis'!
in the preceding section.

In a more recent study of shock and perception, Pustell (1957)
employed geometrical figures (diamond, circle, square, triangle)
rather than taboo words or nonsense syllables. In a pretest phase,
a tachistoscope was used to present cards, on each of which were
mounted three grey forms and one black form. The subjects' task
was to identify the spatial position of the darker figure on each
card. Then in a training phase various subgroups of subjects were
shocked while looking at each of the different black figures.
Tests for the effects of shock were conducted by presenting cards
on which all four figures were black, the requirement now being
to point out the section of the card "which stood out" most.
Because of the brief stimulus exposures, the subjects never realized
that the all-black sets of forms were different from the original
training sets having only one black figure. Significant changes in
perceptual thresholds were obtained with male but not female
subjects. As the result of shock, the males became more "vigilant,"
the particular form previously associated with shock being more

310 THE MOTIVATION OF BEHAVIOR

often denoted as "standing out" than nonshock forms. Women,
however, were affected in the opposite direction, tending to become
"defensive." For them the shock figures were less hkely (not sig-
nificantly) to be seen as "outstanding." Pustell, in attempting to
account for these findings, has postulated that perhaps anxiety
provides a facilitative cue for men which helps them to identify
the shock figure, whereas anxiety (in a way that is not explained)
provides drive for women and thus leads them to be defensive.

The Perception of Socially Taboo Stimuli. Current interest in
the problem of the perception of socially taboo words stems largely
from an investigation by Bruner and Postman (1947), in which
elevated thresholds for taboo as compared with neutral words were
ascribed to perceptual defense. Although the literature on percep-
tual defense is very extensive, explicit definitions of the term are
rare, a fact which may account, in part, for the frequent debates
the field has witnessed in recent years. From implicit statements,
however, it is evident, as we have already indicated, that perceptual
defense means quite different things to different writers. For some
the term is appropriate only to those incidents in which the pre-
perception is both unconscious and specific to the preperceived
stimulus, and in which the defensive reaction affects perception
itself. For other workers perceptual defense seems to denote sim-
ply the empirical finding of higher recognition thresholds for
inimical than for neutral stimuli. And in other instances the term
is used to refer to a more or less conscious avoidance, of the sort
made by individuals when they catch a fleeting glimpse of a grue-
some or revolting scene. Closing the eyes, looking askance, or
thinking of other things are, in this special sense, instances of
perceptual defense. The term perceptual vigilance has also been
variously and vaguely used, though it has received less attention
than the concept of defense, perhaps because vigilance is less
dramatically related to psychoanalytic theory.

During the years since defense and vigilance first came into psy-
chological prominence, two major trends have developed in re-
search and interpretation. The first is characterized by investiga-
tions and explanations designed to prove that the phenomena of
both defense and vigilance can be adequately explained by appeal
to conventional principles of associative learning. The aim of these

' MOTIVATIONAL VARIABLES AND PERCEPTION 311

studies has been to show that differences in the learning history of
subjects prior to their entering the laboratory can account for
what may appear to be laboratory-induced changes in perceptual
processes. In essence, this is simply an associative view of the mech-
anisms mediating either raised or lowered identification thresholds.

The second stream of research arises from the premise that
defensiveness and vigilance are not adjustive reactions of the popu-
lation at large but attributes of individual personalities. On this
view some subjects will be defensive, some vigilant, and others
may be both, depending on the nature of the stimulus material
and the circumstances under which it is presented. Our treatment
of research in this area will involve the consideration of these two
major trends.

Perception of Taboo Materials as an Associative Phenomenon.
The first and most devastating blows against the view that defense
is an unconscious, motivationally determined effect on perception
per se were struck by Howes and Solomon (1950) and by Solomon
and Howes (1951). These authors approached the problem from
a strictly operational point of view, much as we have approached
it in earlier sections of this chapter. This led them to the conclu-
sion that perception is not studied directly, but is inferred from,
or defined by, characteristics of the subject's responses and of the
stimuli presented to him. In the majority of experiments on human
perception, moreover, the subject's responses are verbal or lin-
guistic, and hence it is only when linguistic responses change
that one can properly speak of changes in perception. Thus, to
study perception in these situations is to investigate the manifold
variables of which linguistic behavior is a function. As Solomon
and Howes (1951) have phrased it, ". . . any variable that is a
general property of linguistic responses must also be a property of
any perceptual concept that is based upon those responses" (p.
257).

The principal conception to which this line of attack leads is
that the perception of printed words must depend, in considerable
degree, upon the frequency with which the individual has seen,
pronounced, and used those words in the past. That is, the
stronger the learned associative tendency to give a linguistic re-
sponse to a certain cue, the more readily will impoverished forms

312 THE MOTIVATION OF BEHAVIOR

of that cue elicit the correct response, provided impoverishment
does not itself alter relative associative strengths. In the typical
perceptual experiment, this means that recognition thresholds
should decrease as familiarity increases. In the terminology of this
book, Howes and Solomon were simply asserting that the thresh-
olds for linguistic identifying responses are inversely related to the
habit strengths of those responses.

In criticizing previous studies in which perceptual defense was
alleged to have been operating, Solomon and Howes noted that
the taboo words might have been in many cases less familiar to
college subjects than were the neutral control words. Where this
was true, higher thresholds for taboo words could easily be ex-
plained without invoking the notion of defense. And by a similar
line of reasoning one could also account for instances of perceptual
vigilance, provided only that cogent evidence for the greater
familiarity of the stimuli could be adduced.

This criticism of the concept of perceptual defense has un-
doubtedly had a widespread salutary effect upon research in this
area. Almost without exception, some attempt is made in current
investigations of the phenomenon to equate taboo and neutral
words for familiarity and/or frequency of usage prior to determin-
ing their relative recognition thresholds. Familiarity values are
most commonly allotted to stimulus words on the basis of their
frequency of occurrence in written English, as determined by
Thorndike and Lorge (1944). However, there is some question
whether the Thorndike-Lorge counts are representative of word
frequencies in today's written English. The word "Kotex," for
example, which is sometimes used as a taboo word, does not ap-
pear in these counts. Moreover, it is doubtful whether any index
of frequency based on written English is entirely adequate as an
estimate of word familiarity in the population at large and espe-
cially among college students. As a consequence, those who defend
the concept of perceptual defense (e.g., McGinnies, 1950) can
assert with some confidence that the words whore, bitch, and belly
are far more familiar, at least in spoken English, than the words
beatific, elegies, and vignettes, even though the Thorndike-Lorge
semantic counts show them all to be equally frequent. If the
thresholds for taboo words such as these are elevated, the cham-

MOTIVATIONAL VARIABLES AND PERCEPTION 313

pion of defense may declare that this is in spite of their relatively
high familiarity as estimated, perhaps, from frequency of usage in
bull sessions and bars. And the opponent of defense, relying upon
a different criterion of familiarity, may reason that the elevated
thresholds are due simply to weaker associative strengths. Opin-
ions differ, therefore, with respect to the appropriateness of various
methods of defining familiarity. But it is generally conceded that
any alleged instance of perceptual defense is suspect in the absence
of evidence that the associative strengths of critical and neutral
words have been equated.

Perhaps the best control of the frequency variable has been
achieved in studies where the same words have been used under
conditions designed to make them either threatening or neutral.
McGinnies and Sherman (1952), as we have pointed out, found
that neutral words, when preceded by taboo words, exhibitedr
higher thresholds than when the neutral words were preceded by!
other neutral words. An interpretation in terms of differential!
familiarity cannot reasonably be applied to this experiment though {
residual emotionality might account for the findings. Wienar
(1955) endowed words with either threatening or nonthreatening
characteristics by imbedding them in different contexts for differ-
ent groups. His subjects exhibited sensitization, however, rather
than defense, a threat group requiring fewer recognition trials
than a nonthreat group. This finding is consistent with the sup-
position that threat leads to an increment in drive, which in turn
magnifies tendencies to make appropriate linguistic responses in
the recognition situation.

A survey of experiments in which adequate controls for familiar-
ity were lacking indicates that, with unselected populations, both
higher and lower thresholds are reported for taboo words, the first
result being more frequent than the second. The preponderance
of defensive indications is not great, however, and even if the
familiarity variable were not involved in important degree, proc-
esses other than defense may have been operating. One plausible
explanation, also proposed by Howes and Solomon, is that subjects
may resist saying a taboo word in the presence of an experimenter
even for several trials after the word has been covertly identified.
Thus, for example, a subject might well hesitate to say the word

314 THE MOTIVATION OF BEHAVIOR

penis aloud even though reasonably certain that this specific word
had been presented. Since the tendency to inhibit the overt enun-
ciation of taboo words in certain kinds of social situations is un-
questionably learned, it is not unreasonable to regard this "defen-
sive mechanism" as fundamentally associative.

That response suppression of this kind may often function to
produce "apparent perceptual defense" is indicated by a variety
of studies. Chief among these is the investigation of Whittaker,
Gilchrist, and Fischer (1952), in which it was found that Negro
subjects deliberately refrained from making verbal identifications of
words derogatory to their race until the stimuli became unam-
biguously clear. This conclusion was based on verbal statements
of the subjects and on the fact that the effect was observed when
the experimenter was white but not when he was a Negro. Evi-
dence for comparable effects comes from other studies in which
subjects have reported withholding responses or where significant
subject-experimenter interactions have been observed. The act of
refraining from overt enunciation because of felt or imagined so-
cial threats is itself quite obviously a kind of defense against
anticipated embarrassment. It does not qualify as perceptual de-
fense, however, if by this term one means the selective filtering
of incoming inimical stimulus patterns.

In trying to keep subjects from withholding overt verbal re-
sponses, experimenters have devised a number of different proce-
dures. For example, subjects may be instructed to respond in
writing rather than orally; they may be told ahead of time that
vulgar words are to be presented and that they should not be con-
cerned over speaking them; or they may be given facilitative sets
through instructions which allege that hesitation in uttering the
critical words is a sign of maladjustment and neuroticism. The
results obtained with procedures such as these favor the conclu-
sion that as the tendency to withhold overt responses diminishes,
differences between recognition thresholds for taboo and neutral
words tend to disappear. Postman, Bronson, and Gropper (1953),
for example, whose subjects responded in writing, actually found
taboo words to have lower thresholds than neutral words, a differ-
ence which they attributed to an initial underestimation of the rel-
ative familiarity of the inimical words. Moreover, Freeman (1954)

MOTIVATIONAL VARIABLES AND PERCEPTION 315

reported that when subjects were set to look for taboo words,
recognition thresholds were no different than for neutral words.
It also seems likely that taboo-word thresholds tend to be higher
than neutral-word thresholds at the start of an experimental ses-
sion, the difference disappearing progressively as the subjects come
to expect and get set to report the taboo items (Bitterman and
Kniffen, 1953; Lacy, Lewinger, and Adamson, 1953).

The fact that subjects who are told to expect socially unaccept-
able words tend not to exhibit perceptual defense does not, how-
ever, constitute a crucial argument against perceptual defense.
When subjects are so instructed, they are thereby predisposed to-
ward the making of one or more of a relatively limited number of
responses. This set to respond, which seems to qualify as an asso-
ciative process, might facilitate the identification of critical words.
But if the taboo and neutral thresholds do not differ significantly
under conditions such as these, the advocate for defense could
certainly argue that if the "bad" words had not been selectively
favored their thresholds might well have been higher.

In summary, it can be stated with reasonable confidence that in
many instances where an unconscious mechanism of perceptual
defense is alleged to have been involved, the results can be ex-
plained more simply and more parsimoniously by appeal to well-
known principles of associative learning. Among these principles
are those of frequency of exposure to a stimulus, frequency of
previous elicitation of the response, learned tendencies to with-
hold socially inappropriate responses, and selective sets which
predispose subjects either to look for or not to look for taboo
words. There are few experiments, indeed, in which none of these
associative interpretations has proved reasonable and in which
perceptual defense stands out as the most acceptable hypothesis.

Competing Response Theory of Perceptual Defense. We shall
conclude this chapter shortly with a brief look at studies purport-
ing to show that perceptual defense and/or vigilance are charac-
teristic modes of adjustment of individual personalities. First, how-
ever, let us see whether some of the findings of the experiments
we have just considered can be interpreted in somewhat different
associative terms than those proposed by Solomon and Howes.

Initially, let us assume that printed vulgar words arouse tend-

316

THE MOTIVATION OF BEHAVIOR

encies both to speak and not to speak. Vulgar words (or shocked
words) are thus the counterparts of ambivalent stimuli in a con-
flict situation (cf., e.g., Miller, 1944; Brown, 1957). They tend, as
a result of socially administered reinforcements and punishments,
to elicit incompatible positive and negative responses. Moreover,
since severity of punishment for saying a vulgar word probably
varies directly with its degree of vulgarity, it is further hypothesized
that the tendency for such words to elicit avoidance or nonspeak-
ing responses also varies directly with vulgarity.

These postulated relationships are shown graphically in Fig.
8:12. Here the base line denotes degree of vulgarity, and the
ordinate shows the strength of the positive and negative tend-
encies corresponding to each word. To simplify matters, all words
are regarded as being equally familiar and hence as possessing iden-
tical, strong, positive tendencies. The inhibitory tendency is in-
dicated as weaker than the positive for each word, whether vulgar
or not, since in actual experiments nearly all subjects do speak such
words when they are clearly presented. These assumptions permit
one to deduce, among other things, that overt responses to "dirty"
words should have a longer latency than responses to neutral words,
even when all are presented clearly. To deduce the phenomenon

• .0 •

Positive tendency/

■o

-«'

■•6

•s

^'^^ ^^"~- Negative tendency

1 1 1 1

Low

Moderate
Degree of vulgarity of stimulus words

High

Fig. 8:12. Relative strengths (assumed) of positive tendencies to speak and
negative tendencies not to speak when stimulus words of varying degrees of
vulgarity are clearly presented (i.e., long exposure times are used). The
curves are constructed on the premise that all of the words are equally familiar
(flat positive-tendency curve), but differ in the degree to which they elicit
tendencies to inhibit overt expression.

MOTIVATIONAL VARIABLES AND PERCEPTION

317

of higher tachistoscopic recognition thresholds for taboo words,
however, we need additional concepts.

One principle that seems to hold considerable promise here is
that of stimulus generalization. To apply this principle, let us sup-
pose that the tendency to speak a written word is maximal when
tachistoscopic exposure time is long, but that as the stimulus be-
comes impoverished with reduced exposures, the strength of the
tendency decreases. In short, the function relating exposure time
to strength of reaction tendency is assumed to be similar to a
gradient of stimulus generalization. This supposition is portrayed
graphically in Fig. 8:13, where the upper curve is the strength of
the tendency to speak a moderately vulgar word for each of a num-
ber of different exposure times. (The highest point on this curve
has been set to equal the height of the horizontal positive-tend-
ency curve in Fig. 8:12.) This figure illustrates the further hy-
pothesis that the tendency not to speak the off-color word also
generalizes along the dimension of exposure time. (The maximum
strength of the weak negative tendency equals that for a mod-
erately vulgar word in Fig. 8:12.) Were we to combine this gen-
eralization figure with the previous one, these gradients would ex-
tend out toward the viewer at a right angle from the surface of

MODERATELY VULGAR WORD

Positive tendency.

Short

Moderate
Stimulus exposure duration

Long

Fig. 8:13. Hypothetical generalized tendencies to pronounce and to refrain
from pronouncing a moderately vulgar word. Both tendencies are assumed to
become weaker as the stimulus is impoverished by reducing exposure dura-
tion. The distance (A) between the points of the double-headed arrow denotes
the assumed amount by which the positive tendency must exceed the negative
for the response to become overt.

318 THE MOTIVATION OF BEHAVIOR

Fig. 8:12, Similar pairs of generalization curves could obviously
be constructed for each of the words of Fig. 8:12.

Assume now that the positive and negative tendencies summate
algebraically and that the positive must exceed the negative by a
certain fixed amount for the response to become overt. As has been
noted, in determining thresholds with a tachistoscope, a word is
usually presented for increasingly long exposures until the subject
correctly identifies the word on two or three occasions. In terms of
Fig. 8:13, this means that over successive trials there will be a
progressive increase in the difference between the positive and
negative tendencies. This increase is clearly consistent with the fact
that correct recognition responses become more probable with in-
creased exposure duration. It has been assumed that the identify-
ing response will not be elicited until the positive tendency ex-
ceeds the negative by a certain definite amount, which is indicated
in Fig. 8:13 by the double-headed arrow identified by a delta (A).
For this off-color word, therefore, the recognition threshold should
lie at an exposure duration just slightly greater than one of mod-
erate length.

Suppose, however, that an unusually vulgar word is substituted
for the off-color one. For such a very vulgar word, assuming equal
familiarity, we should expect the positive tendency to be the same
as before but the negative tendency to be greater for all exposure
times. This is schematized in Fig. 8:14, where the negative tend-
ency at maximum is about equal to that for the most vulgar word
in Fig. 8:12. Since the generalized inhibitory tendency is stronger,
as is shown here, for all exposure durations, the positive and nega-
tive curves lie closer together throughout the range of the ex-
posure-duration dimension than was the case for the off -color word.
With the very dirty word, therefore, relatively long exposure dura-
tions must be reached before the difference (A) between the
positive and negative tendencies becomes sufficient to permit the
verbal (recognition) response to occur. As this diagram shows,
the double-headed arrow lies farther to the right than it did in
the previous figure, and since this corresponds to increased ex-
posure durations, a higher threshold for a very dirty than for an
off-color or neutral word can thus be deduced. Of especial sig-
nificance, here, is the fact that neither emotion, nor drive, nor

MOTIVATIONAL VARIABLES AND PERCEPTION

319

VERY VULGAR WORD

Positive tendency

Short

Moderate
Stimulus exposure duration

Fig. 8:14. Hypothetical gradients in the strength of generalized positive and
negative linguistic tendencies for a very vulgar word. Here the difference be-
tween positive and inhibitory tendencies (A) that is arbitrarily assumed to be
required for the evocation of an overt reaction falls at a longer exposure dura-
tion than in the case of a less vulgar word (cf. Fig. 8:13). A higher recogni-
tion threshold for very vulgar than for equally familiar neutral or moderately
vulgar words can thus be deduced.

perceptual defense, nor differential familiarity has been invoked as
a necessary adjunct to the deduction.

Personality Characteristics and the Perception of Taboo Mate-
rials. Those who currently favor the concept of perceptual defense
emphasize the notion that defense is an individual mode of ad-
justment to threat. If their idea is tenable, then an unselected
population of subjects would not be expected to exhibit perceptual
defense. Rather, since some individuals react in a defensive man-
ner and others in a sensitized manner, randomly chosen subjects
should show neither defense nor vigilance, their distinct modes of
reacting being concealed by the process of averaging.

This view, that reactions to threatening stimuli differ consider-
ably from individual to individual and from situation to situation,
leads to research procedures which differ from those we have
previously discussed. The principal change is that subpopulations
of subjects are chosen initially from larger groups by means of
tests purporting to reveal characteristic personality differences.
Perceptual tests are then carried out to determine whether the
personality attributes correlate with degree of defensiveness or
vigilance. In principle this method is like that used in studies of

320 THE MOTIVATION OF BEHAVIOR

need achievement and of manifest anxiety. It is a procedure di-
rected toward the discover)^ of R-R relationships, since it is hoped
that responses made to the items of one test will be correlated with
responses elicited by a second test.

This general approach appears to have stemmed from studies
such as that of Postman, Bruner, and McGinnies (1948), in which
an attempt was made to relate perceptual selectivity to a factor
described as "personal values." By personal values these authors
apparently mean individual differences in the degree to which one
is interested in, or places high value upon certain things or ideas.
Thus a scientist presumably places a higher value upon scientific
principles, relations, and attitudes, than on comparable aspects of
other areas such as religion, economics, or politics. Similarly, the
artist, the politician, and the businessman each has his own char-
acteristic value or interest patterns. And even individuals who can-
not be identified in terms of professional affiliation presumably
have higher values for some areas than for others, and these high-
and low-value areas should be detectable by means of suitable
tests.

In the Postman, Bruner, and McGinnies experiment, college
subjects were tested by means of the Allport- Vernon Study of
Values, a test designed to yield value ratings in areas described as
economic, theoretical, religious, social, esthetic, and political. Each
subject was also given the perceptual task of identifying words in
each of the six value areas. The results of the perceptual test in-
dicated that there was some tendency for the subjects to see words
in their own high-value categories more readily than those in their
low-value groups. Thus an individual with a high religious-value
score was described as having been sensitized to religious words.
Conversely, for such a person^ scientific or theoretical words might
be threatening, and if so, he should exhibit perceptual defense for
such words.

Although some of the statistical and graphic procedures of this
study can be seriously questioned, similar results have been re-
ported by Haigh and Fiske (1952), Solomon and Howes (1951),
and others. It is of interest, therefore, to consider some of the
factors that might produce a positive correlation between test-
defined value orientation and perceptual identifying responses.

MOTIVATIONAL VARIABLES AND PERCEPTION 321

For most writers "value orientation" refers to a kind of inner
personalistic factor capable of organizing both the perceptions and
the overt actions of the individual. Thus, for example, Brown and
Adams (1954) assert that a value area is a "central, cognitive, affec-
tive construct." As such, it is given the properties of determining
both selective responses made to value-test items and selective
perceptions of value-related stimulus words. Postman, Bruner, and
McGinnies say little concerning the nature of value orientation per
se, save that it results from a long process of socialization and
functions to modify perception by means of selective, accentuative,
and fixative mechanisms.

The results of studies relating perception to value orientation
can probably be explained, however, without appealing to value as
such. Thus it might be supposed that a high aesthetic (or other)
value orientation, as defined by scores on the Allport-Vernon test,
means simply that one is interested in, and devotes considerable
attention to aesthetic matters. But a strong interest in aesthetic
and artistic subjects would lead one to read articles and books on
such topics to the relative exclusion of other works. Consequently
the person with a high aesthetic-value score should become more
familiar with the particular words used in writings on aesthetics
than with words in, for example, the areas of science and politics.
If he had such familiarity, then he would more readily identify
aesthetic words when they were presented tachistoscopically than
words of other value areas. And this should hold true even though
the aesthetic and nonaesthetic words occurred with equal fre-
quency in printed English and hence would be defined as "equally
familiar" in terms of a criterion such as the Thorndike-Lorge word
count.

This interpretation, it may be seen, is essentially the associative,
nonmotivational one proposed by Solomon and Howes (1951)
and reiterated by Postman (1953ci), and by Farber (1955). It is
the view that recognition thresholds are determined by familiarity,
that unequal familiarity with words of different value groupings
depends upon one's interests or preferences, and that interests or
preferences might be detected by responses to the items of the
Allport-Vernon scale.
Whether value as a special-purpose cognitive construct need

322 THE MOTIVATION OF BEHAVIOR

be evoked in connection with such an interpretation remains a
matter of individual preference. One could argue, in the interest
of parsimony, that the concept of value is quite superfluous since
its alleged functions as a determinant of behavior can be ex-
plained by assuming that individuals have learned through socially
administered rewards and punishments to be interested in differ-
ent kinds of things. Or it might be argued that value orientation is
somehow an antecedent to, and a determinant of, one's differential
interests, and that these in turn determine familiarity and hence
differential recognition thresholds.

Eriksen (1954) has been an active exponent of the view that
defense and vigilance are idiosyncratic rather than universal modes
of response to threat. He has criticized the conventional proce-
dures in which "dirty" words are presented to randomly selected
subjects on the ground that such words are probably not anxiety
arousing for all or even most typical college subjects. Moreover,
he maintains that an adequate test of the perceptual defense hy-
pothesis necessarily demands some evaluation of the ways in
which individual subjects handle or react to anxiety. On this
view one must demonstrate that individuals differ in degree of
defensiveness to anxiety-arousing stimuli presented in nonpercep-
tual situations. Subjects cannot be said to be perceptual defenders
unless some independent support is thus provided for the assump-
tion that they are generally defensive.

The adoption of this theoretical position generally leads to dis-
tinctive kinds of research. For example, Lazarus, Eriksen, and
Fonda (1951) used, with neurotic subjects, a sentence-completion
test in which some of the sentences could easily be filled out with
sexual or aggressive solutions. Subjects who, on the basis of this
test, could be defined as "expressive" with respect to sex were able
to perceive auditorily presented sexual sentences as well as neutral
sentences. But subjects who tended to block or give distorted
endings to the sexual sentences could not hear the sexual sentences
as well as they could hear the neutral ones. In another study,
Eriksen (1951) has reported that some subjects show signs of
blocking or emotional upset to TAT pictures which usually elicit
aggressive reactions, whereas other subjects express aggression quite
freely in their stories. When tested on a perceptual recognition

MOTIVATIONAL VARIABLES AND PERCEPTION 323

test, the former subjects had higher thresholds for aggressive than
for neutral pictures, whereas the latter subjects had relatively low
thresholds for the aggressive scenes.

Research studies designed along similar lines have been carried
out by Stein (1953), Chodorkoff (1956), and Neel (1954), all of
whom have found defensiveness or vigilance, as revealed by per-
ceptual thresholds, to be correlated with the results of other tests
purporting to identify personality characteristics of individual
subjects.

By way of conclusion it may be noted that although the study of
perceptual defense and/or vigilance as an individual-difference
phenomenon seems like an attempt to prove the reality of percep-
tual defense in the face of such criticisms as those of Solomon and
Howes, it is nevertheless a perfectly acceptable way of proceeding.
The fact that randomly chosen subjects may not exhibit defense,
particularly when frequency of prior experience with the stimuli
has been properly controlled, does not mean that some individuals
might not exhibit the effect. Clearly, one can define defensiveness
in terms of responses to a sentence-completion test or to a test of
some other kind, and then proceed to study the relation of de-
fensiveness thus defined to reactions evoked in other test situations.
The individual-difference method does not, however, preclude the
possibility that the reactions to all tests are due to differential
experiences with the materials of the tests rather than to percep-
tual defense. When individuals who are "expressive" with respect
to sexual matters are selected out of a group, we may simply be
selecting those who, in addition to being expressive, or perhaps
because of their expressiveness, have had more extensive experience
with sexual words, symbols, and ideas. On the other hand, the
test-defined "sexual inhibitor" might well have had fewer exposures
to, and experiences with, materials of a sexual nature. If such
groups differ in their ability to identify sexual words, it might be
more sensible to explain these results by referring to individual
differences in associative strengths rather than to traits of defensive-
ness or to value systems with vague motivational overtones.

324 THE MOTIVATION OF BEHAVIOR

Summary

Of central importance to the discussions in this chapter is the
general problem of the effects of motivational variables upon
perception. However, the perceptions of subjects whose behavior
is being studied by the psychologist cannot be observed directly
by him, and perception must therefore enter into theories of be-
havior in the role of a scientific construct. Like other constructs,
e.g., motivation and associative strength, perception must be ex-
plicitly defined within scientific language, and it must prove its
significance through its capacity to enhance our understanding of
behavior. That a construct of perception may not be needed in
behavior theory is suggested by the fact that many so-called percep-
tual experiments, in which subjects are required to identify or com-
pare physically present stimulus objects, can be successfully con-
ducted and useful stimulus-response laws determined without any
reference to perception. Nevertheless, perception may serve a use-
ful explanatory purpose in some situations, especially those in
which overt identifying responses do not occur until some time
after a stimulus has been withdrawn. Inasmuch as perception is
often defined in terms of subjects' linguistic responses, a motiva-
tional variable may be said to have affected perception when its
introduction leads to changes in the kinds or frequencies of lin-
guistic responses exhibited by subjects in so-called perceptual ex-
periments.

In an effort to relate the motivation-perception problem to
material contained in earlier chapters, an attempt is made to apply
the multiplicative-drive hypothesis to behavior in psychophysical
test situations. From the assumption that empirically determined
response frequencies provide evidence of relative habit strengths
1/ the conclusion is reached that absolute thresholds may either be
elevated or depressed if a motivational variable leads solely to the
arousal of competitive or facilitative habits. But if the variable
affects drive alone, the slope of the psychophysical function should
fcecome steeper, the threshold, as conventionally defined, remain-
ing unaffected. Similar lines of reasoning are applied to the inter-
pretation of data obtained from studies of differential sensitivity
and of recognition thresholds for meaningful stimuli.

MOTIVATIONAL VARIABLES AND PERCEPTION 325

Studies of the effects of primary sources of drive upon behavior
in perceptual situations have yielded suggestive but inconsistent
results. Some observations suggest that food and/or water depriva-
tion enhances the frequency with which need-related words are
given as responses to unstructured stimuli and to near-threshold
presentations of need-related words or pictures. Deprivation op-
erations have also produced contradictory results, both within and
between experiments, and in some cases deprivation leads to
poorer performance even with need-related stimulus items.

A large majority of studies relating motivational variables to
perception consist of those in which the to-be-perceived stimuli
themselves are thought to function as motivational variables. Thus
words that have been associated with positive rewards, with elec-
tric shock in the laboratory, or with social punishments have been
used as to-be-identified stimuli on the assumption that the percep-
tion of such words will be modified by their acquired motivational
properties. Analyzed in this way, the problem may be described as
that of the influence of secondary sources of drive upon perceptual
behavior.

A review of representative investigations purporting to show
that the judged size of objects is affected by their value leads to
the conclusion that the effect at best is slight, especially when the
valued objects are physically present. The judgments of children
and of hypnotized adults appear to be most readily affected by the
variable of object value, and individual differences in the need for
money affect judgments of remembered coin size. It is suggested,
by way of interpretation, that because of socialization processes,-*^
objects of value tend to evoke both stronger instrumental prefer-
ence responses and more frequent verbal responses of the general
class "large" than do comparable nonvalued objects. The elicitation
of more frequent "larger-than" responses by valued objects in a psy-
chophysical investigation is the basic datum indicating that value
leads to an accentuation of judged size.

Interpretations of differences in the recognition thresholds for
emotion-arousing and for neutral words have commonly involved
appeal to mechanisms of perceptual defense or perceptual sensitiza-
tion (vigilance). Defense is usually described as an unconscious
process akin to repression that functions to protect one from

326 THE MOTIVATION OF BEHAVIOR

perceiving things that are distasteful or anxiety-arousing. Early
studies purporting to have demonstrated this phenomenon have
been criticized for failing to equate the familiarity of taboo and
neutral stimuli. Later investigations, in which different criteria of
familiarity have been employed, have yielded inconsistent results,
defense being reported in some cases and vigilance in others. In
most instances interpretive mechanisms other than defense or
vigilance, such as response suppression, familiarity, and predispos-
ing sets, appear to provide more parsimonious interpretations of
lowered or raised recognition thresholds. Higher recognition thresh-
olds for taboo words might also be explained on the grounds that
tendencies both to speak and not to speak such words have been
established through learning and that these tendencies generalize
along the dimension of exposure duration. Assuming that an iden-
tifying response will not become overt until the tendency to speak
exceeds the tendency not to speak by a given fixed amount, it is
concluded that recognition thresholds would fall at longer ex-
posure durations for taboo than for neutral words.

The final section of this chapter contains a brief review of in-
vestigations designed to show that perceptual defense and vigi-
lance are idiosyncratic phenomena. Underlying these studies is
the view that an individual is defensive only with respect to cer-
tain kinds of taboo items and that perceptual defense cannot be
demonstrated, therefore, unless the subjects' defensive areas have
been appropriately identified. Experimental findings have sup-
ported this contention, since defensiveness (or vigilance), as de-
fined by selection procedures of a nonperceptual nature, has been
found to correlate with perceptual thresholds. Even in these in-
stances, however, alternative interpretations couched in terms of
individual differences in familiarity with certain classes of stimuli
merit careful appraisal.

CHAPTER

Miscellaneous
Motivational Problems

This final chapter deals with a variety of problems, which,
though only tenuously related to one another, are all of consider-
able interest to the student of motivation. Each of these topics
might have been woven into the fabric of the preceding chapters,
but only, it was felt, with some loss of continuity and over-all
unity. As we shall see, the bearing of the questions considered
here upon previously discussed problems, though somewhat tan-
gential, is nevertheless significant.

Approach-eliciting Stimuli as Motivational Variables

At several places in the foregoing pages (especially in Chapters
3 and 7) it has been noted that motivational properties are com-
monly attributed to such external stimuli as intense electric shocks,
bright lights, blasts of air, loud noises, and acrid odors. Presum-
ably these stimuli are regarded as motivators because, as experi-
mental variables, they satisfy one or more criteria (cf. Chapter 2)
for the identification of motivational variables. In addition, these

327

328 THE MOTIVATION OF BEHAVIOR

stimuli tend to evoke responses of escape or withdrawal, a charac-
teristic that results in their being described as noxious or aversive.

There are many other external stimuli, however, that tend to
evoke approach rather than escape reactions. These "benign"
stimuli are often described as lures or incentives, suggesting that
they, like their more intense noxious counterparts, may also func-
tion as motivational variables. Since we have not yet considered
this problem in detail, we turn now to the question of whether
and under what conditions approach-eliciting stimulus objects do
indeed serve as motivational variables in addition to fulfilling their
obvious role as behavior directors.

The large majority of instances in which appeals have been
made to the drive-arousing functions of approach-evoking cues
are to be found in studies of so-called exploratory, manipulatory,
and curiosity behavior. Investigators responsible for such studies
have typically postulated the existence of special-purpose drives
that are named in terms of the observed behavior. Thus, explora-
tory, manipulatory, and curiosity drives have been proposed as
explanatory concepts. And in most instances the impression is
conveyed that the external, mild, stimulus objects "pull" the
organism toward them, exemplifying a motivating property that is
"attractive" rather than "repellent." Though the theoretical im-
plications and formal details of this view have not been as care-
fully elaborated as might be desired, the general conception has
proved appealing to those who insist that behavior is motivated as
much by "positive" goals as it is by internal disquietudes.

Existing studies in which concepts of exteroceptively aroused
drives have been invoked may be roughly divided into two groups.
First, there are those in which the behavior-directing goal object
has, in the past, either functioned repeatedly as a reward or has
been closely associated with reward. In these studies, whatever
drive-arousing properties the lure may possess are typically attrib-
uted to learning. In the second group of experiments, the remote
stimulus object is either completely or relatively novel to the
subject; in fact, its capacity to arouse drive is said to be contin-
gent upon its novelty. When such an object is first seen by the
organism, a drive of "curiosity" or "exploration" is allegedly
aroused. This new drive is said to impel the individual to move

MISCELLANEOUS MOTIVATIONAL PROBLEMS 329

toward, and to touch, smell, manipulate, or inspect the object.
In the one case, therefore, drive increments are aroused by highly
familiar objects that have often served as, or have been associated
with, empirically reinforcing conditions; and in the other, en-
tirely unfamiliar sources of stimulation generate special-purpose
drives to explore and/or to manipulate. We consider each of these
possibilities in turn.

Incentive-aroused Drive. The general idea that an event such as
the sight of familiar food may heighten one's motivational level is
by no means recent, but relatively precise, systematic formula-
tions of this idea appear to have been first presented by Spence
(195Itz) and by Hull (1951). As we have already observed (e.g.,
in Chapter 5), these theorists contend that appetitive responses
composing the final eonsummatory chain become classically con-
ditioned to the stimuli provided by food and its environs. Through
stimulus generalization, primarily, certain fractional components
of the eonsummatory response sequence will be elicited even
when the animal is yet some distance from the goal. These reac-
tions (rgs) are thought to produce a drive increment (K) which
combines additively with D and which, like D, multiplies all as-
sociative tendencies. The anticipatory goal reaction (r^) is also
said to produce characteristic interoceptive stimuli (s^s) to which
a variety of responses (either learned or unlearned) may be asso-
ciated. The expectation of food is thus seen as a learned tendency
to make anticipatory, partial responses of the kind involved in
eating. These responses can affect other behavior either motiva-
tionally or associatively or in both ways.

Inasmuch as we have already remarked upon the relation of the
rg-Sg mechanism to the general problem of acquired sources of
drive and to such conceptions as McClelland's affective arousal,
only a few additional comments need be made here. First, propo-
nents of the view that motivation stems from external sources as
well as from bodily disequilibria have largely ignored the possibility
that the rg-Sg-^K mechanism might fulfill some of their needs.
Harlow (1953), for example, in a paper that heartily endorses the
importance of external cues as sources of drive makes no reference
to this mechanism. This omission may be due to his desire to stress
the instinctive aspects of the process as well as to a conviction that

330 THE MOTIVATION OF BEHAVIOR

the so-called secondary sources of drive are of little theoretical
significance. Second, an important implication of the Tg-Sg concep-
tion is that it provides a means whereby the organism's level of
motivation can be enhanced while the goal is being approached,
even though the lure itself remains hidden until the final moment
of goal attainment. One who holds that external stimuli are drive
arousing only while they are actually impinging upon the subject's
distance receptors could scarcely maintain that invisible lures are
motivating. Third, since K, the D-like factor, is the result of con-
ditioning, its magnitude can be relatively independent of the in-
tensity of the conditioned stimulus. The motivational increment is
triggered, as it were, by external stimuli, but its magnitude is
primarily a function of the intensity of the learned internal reac-
tions (rgs) rather than of the intensity of the external stimulus. To
the extent that this may be true, these instances of externally
aroused drive reduce in the final analysis to internal disquietudes
that may be similar to those attending hunger, thirst, and emo-
tionality.

Novel Stimuli as Sources of Drive. The view that unfamiliar
stimulus objects are approached, explored, and manipulated rests
upon a variety of well-authenticated observations. Specifically, it
has been noted that rats and a number of other organisms, includ-
ing man, will move toward some, though by no means all, strange
or novel objects. Upon arrival the organism may touch, push, pull,
sniff at, manipulate, mouth, bite, or lick the previously remote
object. Thus, exploratory, manipulatory, and curiosity behavior
consists of two phases: an initial phase of orienting toward and
approaching a given region or object, and a secondary phase com-
posed of a variety of nonlocomotor reactions that appear after the
novel object has been reached. The reaction patterns of the sec-
ond phase are typically described as "manipulatory," especially
when exhibited by monkeys or men who are capable of com-
plicated manual acts. To the degree that exploration-eliciting ob-
jects are indeed novel, interpretations based on learned, object-
specific, anticipatory responses (e.g., rgS) are patently inappro-
priate. In fact, as familiarity with the novel object increases,
exploration declines. Recourse is sometimes had, therefore, to
the special-purpose concepts of curiosity or exploratory drives.

MISCELLANEOUS MOTIVATIONAL PROBLEMS 331

Experimental Studies of Exploratory and Manipulatory Be-
havior. The rather extensive Hterature descriptive of experiments
on exploration and manipulation has been summarized by Glanzer
(1958), Butler (1958), Berlyne (1958), and Dember and Fowler
(1958). Further detailed review of these studies cannot be under-
taken here, but a knowledge of some of their principal findings is
essential to a further discussion of theoretical issues.

Perhaps the most general statement one can make about these
experiments is that in a wide variety of situations animals often
behave so as to reduce contacts with familiar stimuli or locales and
to increase contacts with novel objects or regions. As an example,
when rats are permitted to run freely through a simple T or Y
maze, they tend, on any given trial, to enter a different arm from
that chosen on the preceding trial. As a result, alternation of sides
is often exhibited on successive trials. Apparently this is not a
result of fatigue or inhibition specific to a particular response, since
the same reaction, e.g., right turning, will be repeated if the stimuli
are changed from trial to trial (Montgomery, 1952). The rat
seems to get tired of looking at the same environment, not tired
of making the same response. The fact that percentage alternation
increases when rats are detained in the goal box of the chosen
T maze arm (Glanzer, 1953), supports this "stimulus satiation"
hypothesis, especially since increases in intertrial intervals pro-
duced by detention outside the maze typically reduce alternation.
Moreover, response alternation is observed whether food is pro-
vided for either choice, or for neither. Also related to the alterna-
tion phenomenon is the finding that rats are more likely to enter
and explore a compartment if it contains a variety of objects
such as wooden blocks than if it contains few or none (Berlyne,
1955).

Not only do animals exhibit preferences for new as opposed to
old environments and sources of stimulation, but the opportunity
to explore new situations appears to have rewarding value for the
learning of instrumental responses. Thus Montgomery (1954)
reports that rats will learn to choose the arm of a Y maze leading
to a large open box containing numerous rectangular blocks (the
Dashiell maze) or, for the same reward, will learn a black- white
discrimination (Montgomery and Segall, 1955). Myers and

332 THE MOTIVATION OF BEHAVIOR

Miller (1954), moreover, found that rats would learn to touch a
door in order to open it to get to the other side of a two-compart-
ment box. And Kish (1955), among others, has shown that rats
will learn to press a lever if the act is followed by an increase in
illumination.

One of the most reliable findings of experiments on exploration
is that with repeated exposures novel objects rather quickly lose
their power to elicit approach and exploration. When periods of
rest are introduced following the "extinction" of exploration, a
kind of "spontaneous recovery" takes place, though recovery is
seldom complete. Moreover, the decrement in exploration result-
ing from frequent presentations of one set of objects or situations
generalizes to other similar objects (Montgomery, 1953).

The most dramatic experiments on the rewarding properties of
visually presented stimuli are probably those of Butler (1953) and
of Butler and Harlow (1954). In these investigations, monkeys
were placed one at a time in a dimly lighted, opaque-walled box.
Two small doors were situated on one wall of the box, and if either
of these was unlatched, the monkey could push it open and look
out into the surrounding monkey-laboratory world. When the
doors were painted yellow and blue, and the blue door was always
unlocked, the monkeys learned to open the blue door and ignore
the other, the only apparent reward for this response being the
opportunity to look through the door for about 30 seconds. This
door-pushing behavior was remarkably persistent, most monkeys
showing a willingness to perform the response for many hours
without flagging (Butler and Harlow, 1954). Harlow (1953) has
suggested that this exemplifies the operation of a "drive to ex-
plore the visual world outside the box" and that the opportunity to
do so is reinforcing. He has also stated that this is an instance of
the arousal of a drive by an external stimulus. In other experiments
Butler (1957) has shown that the frequency of door-pushing re-
sponses increases with hours of deprivation of visual experience,
and that auditory stimuli may also serve as incentives and/or
rewards.

Manipulatory activity is similar, in certain respects, to the kinds
of exploratory behavior we have just been considering. If sticks,
mechanical puzzles, or other manipulatable objects are presented

MISCELLANEOUS MOTIVATIONAL PROBLEMS 333

to a monkey, it will sooner or later approach and handle them. If
the object consists of several separable components, the monkev
will eventually disassemble it and in this way "solve" the problem.
Moreover, as with visual exploratory actions, these disassembling
reactions can be elicited over and over again (Harlow, 1953) . Typi-
cally, none of the conventional appetitive or social rewards is ad-
ministered following a successful puzzle-disassembling response,
and yet proficiency in this activity may increase significantly over
trials. The persistence of the behavior and its improvement with
practice are considered to support the contentions that the animal
has a manipulatory drive and is rewarded by performing manipula-
tions. The greater perseveration of manipulatory activities in
monkeys, as compared with exploratory behavior in rats, may be
ascribed to species differences or perhaps to the greater reinforcing
power of manual manipulation as compared with mere visual in-
spection.

That the research findings are not always as consistent as the
foregoing experiments might suggest is indicated by reports of oc-
casional failures to obtain approach and exploration with respect
to strange objects. Young chimpanzees, for instance, do not look
at novel objects and do not approach and explore them until they
have been exposed to such objects for prolonged periods (Welker,
1956). Rats also resist leaving the home cage in order to explore
an elevated runway (Montgomery, 1955), and monkeys will not
learn to open a door to look out at a large dog or to hear recorded
vocalizations of a monkey colony whose members are being
molested (Butler, 1958). Some stimuli, therefore, even though un-
familiar, do not elicit exploratory behavior or function as rein-
forcers for the acquisition of new responses. In some experiments
repeated presentations of such avoidance-eliciting objects are ac-
companied by an initial, progressive increase in exploration, fol-
lowed by a terminal decline. This has been interpreted to mean
that exploration increases as fear is extinguished, with exploration
then declining in the usual manner as stimulus novelty wears off.

Though advocates of the concepts of exploratory and curiosity
drives have pressed their views with vigor and enthusiasm, numer-
ous problems remain to be solved before their interpretations can
be said to carry complete conviction. For one thing, the conditions

334 THE MOTIVATION OF BEHAVIOR

under which these alleged special drives are aroused have not been
carefully specified. As a consequence, one cannot be certain which
bits of behavior require the postulation of these special-purpose
drives and which do not. The topographical properties of the
so-called exploratory behaviors and of other, nonexploratory activi-
ties are in many respects identical. For example, if a rat moves
from one end of a box to the other, one has no way of knowing,
from the behavior alone, whether the rat is motivated by the
sight of the other end of the box, by dissatisfaction with the end
where he is, or simply by a tendency to become active. Moreover,
as Berlyne (1958) has pointed out, a specific behavior pattern may
be scored as exploratory if it is exhibited in one part of the test
apparatus but not if it appears in another. If a rat runs down an
alley and sniffs at an object at the end, he is usually said to be
exploring, but if he stays in the starting compartment and sniffs at
the walls, he is not exploring (by the usual criteria) since he hasn't
gone anywhere.

Most frequently the label "exploratory" is applied when an
animal does something in a situation where he receives no obvious
reward for his performance. But the presence or absence of an ex-
ploratory drive is not announced until after the behavior has been
observed to occur. If a strange object is approached, curiosity drive
is invoked; but if the same object is not approached, the existence
of a fear drive is asserted. Thus the presence of a drive to explore
is sometimes inferred from, and at the same time used to explain,
behavior of moving from one place to another, especially if there
is no other apparent reason for the movement. The postulation of
an exploratory drive in this way is quite circular, and therefore of
questionable worth as a scientific explanation. Not all assertions
concerning the presence of exploratory or curiosity drives are as
clearly circular as this, but the problem has not been squarely
faced nor adequately solved.

Another difficulty arises when one attempts to define stimulus
novelty. Is a block of wood a novel object for the rat? If so, how
novel? Probably if a rat has been raised in the usual way in the
laboratory, it has seen other objects resembling, to a greater or
less degree, a block of wood. Pellets of laboratory chow are often
fashioned in rectangular form, and their color is similar to that of

MISCELLANEOUS MOTIVATIONAL PROBLEMS 335

natural wood. So when a rat approaches and sniffs at a small block
of wood in a maze, how can we be certain that this is not an
instance of transfer of training, mediated by stimulus similarity? If
so, need we worry either about stimulus novelty or about the
curiosity drive that novelty is alleged to arouse? As with explora-
tory drive, the specification of stimulus novelty rests, to an unsatis-
factory degree, upon the observation that the expected behavior
has indeed occurred. Novel objects are supposed to arouse a drive
to explore such objects, but degree of novelty is typically not
evaluated independently of exploration.

Of extant theories of exploratory and curiosity behavior, Ber-
lyne's (1955) is probably the most carefully and precisely struc-
tured. According to this theory, which in its formal properties
resembles the Hull-Spence r^-Sg formulation, when novel stimuli
fall upon an organism's receptors, a drive-stimulus-producing re-
sponse labeled "curiosity" will be evoked. The "drive-to-explore"
or the drive of "perceptual curiosity" acts, apparently, to facilitate
behavior of exploring the objects that generate the response-pro-
duced drive. Exploratory activities reduce the drive to explore, and
when this occurs following a response, that response will tend to
be learned. In addition to functioning as a reinforcing agent, ex-
posures to novel stimuli through exploratory activities, bring about
a weakening of the curiosity drive. According to Berlyne, the proc-
ess of curiosity diminution can be described by Hull's model for
the extinction of conditioned responses, a model in which perma-
nent decrements are explained by the concept of conditioned in-
hibition and temporary diminutions by reactive inhibition.

Because of the similarity of this conception to the Ty-Sy expect-
ancy mechanism, it is of interest to compare the two in more de-
tail. First, whereas the Tg-Sg configuration clearly refers to a learned
reaction, the drive-producing curiosity response must be instinc-
tive, since only novel stimuli can arouse curiosity, and only objects
that the organism has never seen are really novel. Second, although
expectancy-produced drive becomes stronger as a function of the
number of encounters with a rewarding object, curiosity drive is
weakened by the same variable. Repeated experiences with food
that is assumed to be drive reducing strengthen the expectancy
mechanism; but repeated exposures to a novel stimulus, which in

336 THE MOTIVATION OF BEHAVIOR

a comparable manner reduces the curiosity drive, weaken the power
of the stimulus to arouse the drive. Conversely, anticipatory goal
responses are extinguished when the rewarding object is omitted,
while curiosity is extinguished when the rewarding object is kept
in the situation.

Although Berlyne has tried to cast his views into a form consist-
ent with Hullian theory and to extend it into the general area of
perception, his usage of the term drive is not coordinate with
Hull's. Berlyne's curiosity drive is apparently a drive to perform a
specific act or a specific group of acts. Conceived in this manner,
it is a special-purpose drive incorporating directive properties, and
is therefore not the same as Hull's general D. Perhaps Berlyne
adopted this position because one cannot explain why a subject ap-
proaches a novel object instead of doing something else, if one
assumes that the sight of the object leads only to heightened D.
An increase in D does not activate or multiply approach tendencies
alone, and reactions of exploring and manipulating would not be
facilitated by intensified D unless these responses were, for other
reasons, the dominant members of the response hierarchy. Never-
theless, in the typical exploratory situation, approaching may often
be the dominant response. An animal, when placed into a maze
at the starting point, obviously cannot back up. If he moves at all
it must be away from his initial position and therefore toward the
remotely located, to-be-explored objects.

Another account of exploratory behavior requiring analysis is
that proposed by Butler and Harlow in conjunction with their
experiments on the "visual-exploration drive." According to these
writers, visual exploratory responses are members of a broad family
of curiosity or exploratory drives (including manipulatory drives)
that are elicited by external stimuli. This interpretation has been
suggested as an explanation of the behavior of the monkeys who
learned new discriminatory reactions for visual rewards.

That this view leads to difficulties should be clear from the
following: If the act of looking out of the window at the outside
(laboratory) world is the specific event that arouses the visual-ex-
ploration drive, then the monkeys do not have this drive until they
have succeeded in opening the correct window. The drive aroused
by the final act of visual exploration cannot, therefore, provide

MISCELLANEOUS MOTIVATIONAL PROBLEMS 337

the impetus for the response of window opening. External stimuh,
when seen, may indeed incite a drive to explore with the eyes, but
it is illogical to say that those stimuli, while they are invisible to
the monkey, actively function to arouse the drive to work in
order to see.

But the panel-pushing and latch-manipulating activities of
Butler's monkeys were presumably motivated in some way, and
the motivating agency must have been functioning prior to the
time the monkeys obtained their visual rewards. Several possibil-
ities suggest themselves. For one, being unable to see may be the
crucial variable. This possibility is supported by Butler's study
(1957), in which visual exploratory responses were found to in-
crease in frequency as a function of the amount of visual depriva-
tion. Or since, in the typical experimental setup, auditory stimuli
are not excluded, and the confined monkey can hear the sounds
produced by other members of the colony and by numerous people
in the laboratory, these noises could serve as drive-producing in-
centive stimuli. Still another view is that confinement in an opaque
box is a mild anxiety-inducing variable for the monkey (Brown,
1953Z)). When coupled with the auditory cues, this might be
sufficiently motivating to explain the observed behavior. And
finally, it might even be supposed that if seeing is rewarding, for
whatever reasons, then eye movements and other components of
the sequence of looking responses could become conditioned to
environmental stimuli in much the same way as consummatory
responses. Were this the case, then generalized, fractional, anticipa-
tory seeing responses (rg-Sg) might be elicited in visually restrictive
environments and, like the analogous rg-Sg mechanism, might add
an increment to drive. Perhaps in this way one could speak more
precisely of the motivating properties of the "expectation of see-
ing."

Harlow (1953) has maintained that exploratory and manipula-
tory activities weigh strongly against a drive-reduction theory of
reinforcement, since these actions do not produce immediate
diminutions in any of the primary physiological needs. There is
no existing experiment, however, in which the possibility that
some kind of drive reduction has occurred has been completely
ruled out. Thus if a monkey becomes anxious when confined in a

338 THE MOTIVATION OF BEHAVIOR

dimly lighted box, a reinforcing reduction in anxiety may occur
when it gets a chance to see other monke^^s and/or the familiar
laboratory environment. Supporting this view are the previously
cited observations that monkeys will not learn responses that are
followed by such stimuli as the sight of a large dog or the sounds
of other monkeys being molested (Butler, 1958). It is also ap-
parent, in this connection, that if a confined monkey has a drive
to see, as Harlow asserts, then the act of seeing must reduce this
drive. One might thus be led, oddly enough, to a drive-reduction
theory, albeit not a need-reduction theory, of the reinforcing effects
of visual exploration. On these grounds the external stimulus be-
comes a drive-to-see reducer, not a drive-to-see arouser. Incidentally,
Berlyne holds that external cues play both of these roles, arousing
a drive when first seen and reducing that drive with further see-
ing.

It seems justifiable to conclude from this analysis that while
animals do indeed engage in activities of approaching and explor-
ing biologically neutral objects, it is far from certain that a special-
purpose exploratory drive need be invoked to explain such activi-
ties. Moreover, though monkeys and other animals do learn new
responses for rewards of visual stimulation, the nature of the mo-
tivation underlying the evocation of these responses remains un-
clear. Before the visually rewarding events are seen, motivational
processes other than those aroused by to-be-seen stimuli must be in
operation. Apprehension due to isolation and confinement, an-
ticipation of seeing, or auditory stimuli may serve in this role as
motivators. Lastly, although there are many reports of instances
in which animals will learn to solve a T maze or a discrimination
problem for an exploratory reward, alternative explanations have
seldom been carefully ruled out. Thus the presentation of novel,
reinforcing stimuli has typically been confounded with an increase
in space (cf. Montgomery, 1954), and hence the reinforcement
may have been provided by escape from confinement or by the
opportunity to run, which is known to be reinforcing (Kagan and
Berkun, 1954), rather than by the sight of unfamiliar objects.
Experimenters such as Chapman and Levy (1957) have con-
sidered this problem and have concluded that novelty may be
reinforcing even when increased space is controlled, although the

MISCELLANEOUS MOTIVATIONAL PROBLEMS 339

effect disappears when tlie animals are made hungry. Consider-
able research remains to be done, therefore, before we can be
entirely certain that novel stimuli per se are reinforcing and that
no negative factors such as confinement, anxiety, or stimulus
deprivation are also involved. Nearly all currently available studies
can be interpreted as supporting either the hypothesis that the
animals are tired of looking at one portion of their world (stimulus
satiation) and hence are relieved to get out, or that they are
excited and pleased (curiosity drive) to see something new.

On the Conditioning of Emotional Responses

As we have noted, principally in Chapter 5, there is reason to
suppose that fears and other emotional reactions can become as-
sociated, through learning, with biologically neutral stimuli and
may thereby come to serve as acquired sources of drive. But the
problem of how such emotional responses are learned has been
postponed to the present chapter. Since most of the written mate-
rial on this topic deals with the learning of fear, we shall restrict
our discussion to this emotion, omitting the problem of whether
positive affective responses are acquired by processes other than
those involved in fear conditioning.

Theories of Fear Conditioning. Theorists who have concerned
themselves with the question of how emotional reactions are
learned may be divided into two broad groups on the basis of the
answers they have suggested. One group consists of those who hold
that the acquisition of conditioned emotional responses is gov-
erned by the same processes that control the learning of nonemo-
tional responses. The members of the second group maintain that
different processes control the learning of the two kinds of reac-
tions. The members of both groups, however, disagree among
themselves as to which mechanisms should be assigned positions
of central importance.

Probably the best-known members of the first group are Guthrie
(1935) and Hull (1943, 1951, 1952). According to Guthrie, an
association is formed between a stimulus and a response, whether
the response is a skeletal movement or an internal emotional activ-
ity, whenever the response occurs in the presence of the stimulus.

340 THE MOTIVATION OF BEHAVIOR

The mere contiguity, or the occurrence together in time, of stimulus
and response is sufficient to produce the associative bonding that is
the essence of learning. For Hull also, stimuli and responses must
be contiguous if stimuli are to acquire tendencies to elicit reac-
tions. But in addition to contiguity, a reduction in drive following
the elicitation of the response is regarded by Hull as an essential
condition for the formation of associations. During the condition-
ing of fear, the noxious UCS is assumed to serve as a primary
source of drive, and its termination, therefore, constitutes a reduc-
tion in drive. Both Hull and Guthrie thus agree that contiguity
is a necessary condition for the growth of associations. They dis-
agree in that Hull considers drive reduction to be an additional
necessary condition, whereas Guthrie does not. As members of the
"same-processes" group, these theorists believe that the learning
both of emotional and nonemotional responses can be explained
in terms of one set of principles.

Notable members of the second, or "different-processes," group
are Schlosberg (1937), Skinner (1938), and Mowrer (1947). For
Mowrer, whose views we have chosen to examine in detail, S-R
contiguity is both the necessary and the sufficient condition for
the learning of emotional responses. Instrumental reactions in-
volving skeletal muscles are not learned, however, unless a reduc-
tion in drive follows such acts. The fear-conditioning aspect of
Mowrer's interpretation, which in general closely resembles the
views of Schlosberg and Skinner, is thus seen to be essentially
Guthrian, whereas Mowrer's position with respect to instrumental
learning is identical with Hull's. Although Spence (1956) has not
seriously concerned himself with this problem, he has tentatively
proposed a different-process theory that is the exact opposite
of Mowrer's. According to Spence, instrumental actions are per-
haps learned through the operation of contiguity alone, whereas
the acquisition of autonomically controlled reactions may rest
upon the occurrence of drive reduction in addition to contiguity.

Apparently, then, learning theorists, at least those who regard
learning as involving associations between responses and stimuli,
concur in the belief that contiguity is an indispensable condition
for the acquisition of emotional reactions. They do not agree, how-
ever, as to whether additional variables, particularly those upon

MISCELLANEOUS MOTIVATIONAL PROBLEMS 341

which the immediacy or completeness of drive reduction might
depend, are also necessary. Thus the issue is reduced to this basic
question: What empirical variables determine the conditioning
of emotional responses?

Those who, like Mowrer, hold to a strict contiguity view of emo-
tional conditioning would probably maintain that the following
experimental variables are of primary importance for the growth of
emotional associative strengths: (1) the interval between the
onset of the CS and the onset of the UCS (degree of contiguity);
(2) the intensities of the CS and the UCS; and (3) the frequency
and recency of S-R pairings. Presumably, any variable that might
affect the time of occurence, or the abruptness or the complete-
ness of the drive-reduction process would be rejected as of no sig-
nificance for emotional conditioning. One such variable is UCS
duration. If drive reduction occurs when the UCS is terminated
and if drive reduction is reinforcing, then the more protracted
the UCS the greater the delay of reinforcement. From the view-
point of a contiguity theorist, however, varying the interval be-
tween the emotional response and a reduction in the drive, other
things constant, should have no effect upon the strength of the
conditioned emotional associative tendency.

Theorists who do not hold that mere contiguity is a sufficient
condition for the formation of emotional associations would
nevertheless grant the importance of the above variables of con-
tiguity, stimulus intensity, frequency, and recency. But in addition
they would maintain that variables such as UCS duration, which
might affect the drive-reduction process, should not be dismissed
until careful experiments have shown them to be irrelevant or in-
significant.

If isolating significant variables lies at the heart of the problem
of emotional conditioning, then it would appear that definitive
answers could be obtained from straightforward experimental in-
vestigations. But such inquiries could provide us with a complete
list of relevant variables without resolving the issue. Those con-
cerned with this problem might agree on a group of variables and
yet disagree as to the nature of the mechanisms activated by those
variables. For example, the time at which an electric shock is
terminated might be shown to affect emotional conditioning, but

342 THE MOTIVATION OF BEHAVIOR

the question could still be raised as to whether drive reduction does
or does not accompany shock termination. Perhaps for the
masochist, shock cessation is drive arousing rather than drive re-
ducing. Similarly, an increase in the duration of an electric shock
may result in augmented pain, or, if adaptation occurs, in dimin-
ished pain. And the variable of degree of S-R contiguity may affect
learning through any one of a variety of mechanisms of which we
know nothing as yet.

Mowrer, for example, made no distinction between empirical
variables and the hypothetical mechanisms underlying those vari-
ables. Consequently, in rejecting the assumption that the mech-
anism of drive reduction was necessary for emotional conditioning,
he also rejected all of the empirical variables on which drive reduc-
tion might depend. This he need not have done. Like Guthrie, he
could have maintained that certain of the "law-of-effect" or drive-
reduction variables influence the strength of an emotional response
by protecting associations from unlearning rather than by strength-
ening them in the first place. By this stratagem Mowrer could have
retained any of the empirical drive-reduction variables while reject-
ing the drive-reduction mechanism.

Illustrative Investigations. Concerning experimental attempts
to determine whether contiguity alone is sufficient for the learn-
ing of fear reactions, none has thus far yielded unambiguous re-
sults. This is a difficult field, however, in which to design crucial
experiments, since none of the alternative hypotheses has been
stated with sufficient precision to make exact tests possible.

To illustrate the vexing nature of the problem, let us examine
the procedures used in studies purporting to bear on the issue.
One of the experimental designs often used to evaluate the rela-
tive merits of the contiguity-alone versus the contiguity-plus-drive-
reduction conceptions is shown in Fig. 9:1. Here it will be seen
that the classical aversive conditioning procedures used with the
two groups are identical in all respects save that the duration of
the UCS is longer in the case of group II than in the case of group
I. That is to say, the CSs are identical, and the contiguity variable
is also presumably identical, since, with the CS-UCS intervals
equal, the time between the CS and the unconditioned response
must also be the same for the two groups.

MISCELLANEOUS MOTIVATIONAL PROBLEMS 343

Now on a drive-reduction hypothesis it might be predicted that
the members of group I would show more fear conditioning than
the members of group II because the drive-reducing event (shock
cessation) occurs more immediately in group I than in group II.
The assumption that associative strength varies inversely with delay
of reinforcement is, of course, also involved in this prediction.
Thus, if it can be shown that a short-shock group is no more
fearful than a long-shock group, this evidence would weigh
against Hull's delay-of-reinforcement hypothesis and against the
drive-reduction view. But this conclusion would be justified only if
one could be sure that the delay-of-reinforcement variable had not
been overridden by other factors. One such factor is shock in-
tensity. Thus, if shock duration is permitted to vary while delay-
of-reinforcement is being manipulated, this variation may lead
to changes in effective shock strength. As common sense would
seem to suggest, a CS that signals the coming of a long shock
might well arouse more fear than one that is premonitory of brief
shock (cf. Miller and Dollard, 1941, p. 58). Perhaps the uncon-
ditioned pain is more intense with long shock, even with physical
intensity held constant, or the drive reduction occasioned by the
cessation of a longer shock may be greater than that following the
offset of brief shock. If a short-shock group shows more condi-
tioned fear than a long-shock group, the drive-reduction theorist
may feel encouraged. But his theory is not refuted even if a sig-
nificant difference in the reverse direction is obtained. If long

On Off

CS I I

GROUP On Off

I l/CS- Shock I I

On Off

CS I I

GROUP On Off

n UCS-ShocW I

Time

Fig. 9:1. A schematic representation of an experimental arrangement com-
monly used to determine whether the strength of a classically conditioned
response varies with the duration of the UCS.

344 THE MOTIVATION OF BEHAVIOR

shock leads to more fear than short shock, this effect may mean
that events associated with the termination of shock are affecting
the process of emotional conditioning. And if they are, then fear
conditioning may be a function of other variables than the mere
contiguity of the CS and the unconditioned response.

Experimental studies employing designs much like that in Fig.
9:1 have been reported by Sullivan (1950), Bitterman, Reed, and
Krauskopf (1952), Mowrer and Solomon (1954), and Zeaman and
Wegner (1954). Sullivan's experiment, involving the conditioning
of the GSR in human subjects, with short and long intense audi-
tory stimuli as UCSs, provided suggestive but not conclusive
evidence that the short-duration stimulus was more effective in
producing conditioned GSRs. Bitterman et al. attempted to condi-
tion GSRs by means of short and long shocks. While the long-
shock procedure tended to yield responses of larger amplitude than
the short shock, the differences fell far short of statistical signifi-
cance. Unfortunately, the subjects in this study served as their
own controls, the short shock being paired with the lighting of
one 15-watt lamp on one half of the 16 trials and the long shock
being paired with the lighting of an identical lamp on the other
half of the trials. Hence, the failure to obtain differences in either
direction, in spite of the fact that all subjects whose records were
used said they knew one light was followed by long and the other
by short shock, is not surprising. In the Mowrer-Solomon experi-
ment, rats given fear-conditioning training with 10-second shock
showed some indication (10 per cent level of confidence) of being
more fearful than rats for whom the shock duration was only 3
seconds. On the basis of this evidence Mowrer and Solomon sug-
gest that if the short shock had been as brief as 0.5 second, "...
it is virtually certain that it would be reliably less effective in
producing fear than would a shock of long duration" (p. 21). This
is consistent with the intuitive expectation that cues denoting the
coming of a short shock should be less fear arousing than cues
that forewarn of long shock. Nevertheless, Mowrer and Solomon
still maintain that these results support their views that the events
attending the termination of a UCS have no effect upon fear learn-
ing. This conclusion is puzzling, since if one group does exhibit
more fear than the other, with S-R contiguity equated, then, con-

MISCELLANEOUS MOTIVATIONAL PROBLEMS 345

trary to Mowrer's interpretation, contiguity is not the whole story.
If genuine differences in fear can be produced by the use of long
and short shocks, then the variable of UCS duration, involving as
it does the time at which the noxious stimulus goes off, may in-
deed have something to do with fear learning.

Zeaman and Wegner, in a study of conditioned cardiac re-
sponses, have found that a relatively short shock of 2 seconds dura-
tion leads to conditioned acceleration during the CS-UCS interval.
With a 6-second shock, however, the conditioned response is a
deceleration in heart rate. This difference in the form of the con-
ditioned responses is ascribed to the fact that the heart is accelerat-
ing at the time of shock reduction in the short-shock condition,
but is decelerating at the time the long shock terminates. The
authors interpret their results as supporting the view that drive
reduction is the mechanism of conditioning in the case of this
autonomically mediated response.

In connection with designs of the type illustrated in Fig. 9:1, it
should be noted that, with the exception of Sullivan's experiment,
apparently none of the investigations has been conducted with an
optimal CS-UCS interval. The work of White and Schlosberg
(1952) and of Moeller (1954) indicates that a CS-UCS interval
of about 0.5 second yields optimal GSR conditioning. In the study
of Bitterman et al., this interval was set at 5 seconds. According
to White and Schlosberg, this should yield no conditioning at all,
and some question can be raised whether in fact Bitterman et al.
did indeed obtain conditioned responses, since their design in-
cluded no controls for pseudoconditioning. It may also be ob-
served, as Miller (1951) has pointed out, that it is not always cer-
tain, when shock duration is being varied in animal experiments,
that drive reduction occurs only at the instant when the experi-
menter terminates the shock. If an animal is free to move about on
an electrified grid, the shock may actually be intermittent, for the
rat can momentarily escape shock by leaping and running. This
could mean that a long shock, in effect, is a series of short shocks
involving multiple drive reductions and reinforcements. In rat
experiments some control over this factor has been provided by
the use of shock compartments with low ceilings, but so long as
the animals are free to move and the shock electrodes are not fixed

346 THE MOTIVATION OF BEHAVIOR

to the animal, the control may be less than perfect. In any event,
very carefully designed and well-controlled experiments will be
needed before we can be sure that long shocks do not produce a
degree of fear different from that produced by short shocks, and
that delay of reinforcement is not a significant factor.

An investigation by Mowrer and Aiken (1954) provides an addi-
tional example of problems attending the design of crucial experi-
ments on the contiguity-alone versus the contiguity-plus-drive-
reduction views. Figure 9:2 shows the various fear-conditioning
procedures used by Mowrer and Aiken. As this figure indicates,
UCS duration was the same for all groups, but the position of the
CS with respect to the UCS was varied progressively from group
I through group IV. In a fifth group, not shown in the figure, the
CS was withheld until two minutes after the UCS terminated.

In explaining this experimental design, Mowrer and Aiken state
that according to Hull, the nearer the CS is in time to the termina-
tion of the UCS, the greater should be the capacity of the CS to

GROUP

I UCS

GROUP

n UCS

GROUP

m UCS

CS \ L_

GROUP

IV UCS

Time

Fig. 9:2. Four different experimental paradigms used in an experiment de-
signed to evaluate the relative merits of the contiguity-alone versus the
contiguity-plus-drive-reduction views of emotional conditioning. {Adapted
from Mowrer and Aiken, J 954.)

MISCELLANEOUS MOTIVATIONAL PROBLEMS 347

evoke fear. The alleged expectation from Hull's views, therefore, is
that fear should be conditioned more strongly in group III, say,
than in group I. Unfortunately, this is a misinterpretation of Hull's
position, since for him the interval of importance for fear con-
ditioning is not the time between the CS and the offset of the
UCS. Rather, Hull holds that the significant intervals are ( 1 ) that
between the CS and the response, and (2) that between the re-
sponse and the reinforcing state of affairs. In the Mowrer-Aiken
design, if it is assumed that the response is always aroused shortly
after the onset of the UCS, then, since the UCS duration is held
constant, the interval between response and reinforcement is also
constant. However, as the position of the CS relative to the UCS
is progressively shifted from that used with group I to that used
with groups IV and V, there is clearly a change in the temporal
separation of CS and response. But on Hull's views, this change
in the position of the CS would be expected to lead to poorer and
poorer conditioning, since it involves a change from a forward
conditioning procedure to an extreme degree of backward condi-
tioning. It is not surprising, therefore, that Mowrer and Aiken
found group I to be the most fearful, with fear becoming progres-
sively less in the groups for whom the procedure became more and
more "backward." It is well known to students of conditioning
that backward conditioning arrangements yield little if any stable
conditioned responses. Davitz (1955), using stimulus presentation
orders like those of Mowrer and Aiken's groups I and III, also
reports that more fear is conditioned when the CS precedes the
onset of the UCS than when it coincides with UCS termination.
However, Davitz also fails to note that the manipulation of the
position of the CS relative to the UCS involves a shift from
forward to backward conditioning. Strouthes and Hamilton (1959)
have also employed experimental procedures like those used with
groups I and II of the Mowrer-Aiken study. Their animals, how-
ever, showed enhanced bar pressing during the presentation of
the fear-arousing CS rather than inhibition, as in the Mowrer-Aiken
investigation. But Strouthes and Hamilton's group-II animals were
more fearful than their group-I subjects, as estimated by degree
of bar-pressing facilitation. Arguing from the position that delay
of reinforcement is shorter for group II than for group I, and

348 THE MOTIVATION OF BEHAVIOR

from the further assumption that the best index of the delay
interval is provided by the time from CS onset to UCS termina-
tion, Strouthes and Hamilton conclude that their results tend to
support the drive-reduction position. Finally, Gerall, Sampson,
and Boslov (1957) have clearly shown that a known reinforcer
(electric shock) affects the conditioning of pupillary dilation,
which is a response controlled by the autonomic nervous system.
Light offset was found to be an adequate UCS for the evocation
of the pupillary response, but no conditioning occurred when the
CS was paired with light offset alone. However, when the UCS
consisted of shock paired with light offset, or shock alone, condi-
tioning of pupillary dilation was readily obtained.

Physiological Studies Related to the Concept
of General Drive

When Hull (1943) outlined his concept of a nonspecific drive
(D), his speculations rested entirely upon observations of gross,
molar behavior. At that time there were scant grounds for sup-
posing that a general-purpose motivating function of this sort
could be meaningfully ascribed to the activities of any particular
neural system or group of systems. In recent years, however, this
picture has been altered as a result of the discovery of new facts
concerning the physiological and neurological mechanisms of
motivation. Of special relevance for the hypothesis of nonspecific
drive are data that have been interpreted to mean that the brain-
stem reticular system has widespread motivational functions. Addi-
tional findings, particularly those from studies of intracranial elec-
trical stimulation, have shed new light on such matters as the
drive-reduction theory of reinforcement and the tendencies of
organisms to avoid some stimuli and approach others. We shall
discuss these several matters in the sections to follow.

Drivelike Functions of Reticular System Activity. A detailed
elaboration of the anatomical minutiae of the reticular formation
is clearly beyond the scope of this book, and for such information
the reader must consult other sources (e.g., Rossi and Zanchetti,
1957, and Jasper, 1958). In broad outline, however, the brain-
stem reticular formation consists, according to Lindsley (1957),

MISCELLANEOUS MOTIVATIONAL PROBLEMS 349

". . . of a rather dense network of neurons which forms a kind of
central core extending from the medulla of the lower brain stem
to the thalamus in the diencephalon. It extends through the regions
of the pons and the midbrain tegmentum upward through the
caudal portions of the hypothalamus and subthalamus" (p. 57).
Activity in the brain-stem reticular system can affect the function
of structures lying both below and above. In particular, fibers
passing downward from the reticular formation into the spinal
cord carry impulses that assist in the regulation (either by inhibi-
tion or facilitation) of complex postural and phasic muscular
reactions Additionally, and perhaps more importantly for our
purposes, other fiber systems composing the so-called ascending
reticular activating system pass upward to basal ganglia, thalamus,
hypothalamus, hyppocampus, and cortex.

As a consequence, apparently, of the anatomical diffusion of
these ascending reticular system fibers, impulses from lower seg-
ments of the reticular formation can bring about widespread
changes in cortical activity. This "activation effect," as it has
been called, which was first reported by Moruzzi and Magoun
(1949), is a modification of the waveform pattern of the electrical
potentials recorded from the brain. When a subject is asleep, the
form of these electrical impulses differs markedly from that charac-
teristic of waking or excited states. Electrical stimulation in the
region of the reticular system was found by Moruzzi and Magoun
to change the cortical brain wave from the "sleep" pattern to the
"waking" or "excited" form. Of possible significance, for theories
of motivation, is the fact that this "electrocortical activation" is
widespread, being observed at numerous points on the cortical
surface. Moreover, as Lindsley, Bowden, and Magoun (1949) have
shown, a progressive decline in amount of "electrocortical activa-
tion" is associated with increasingly widespread destruction of the
reticular system. At the very least, then, the nonspecific activating
effects of ascending reticular system activity appear to be con-
sistent with the construct of a general-purpose drive.

That the arousal functions of the ascending reticular formation
are analogous to, if not completely commensurate with, the notion
of a nonspecific drive was apparently first noted by Hebb (1955),
though Lindsley (1951) had seen the broad implications of reti-

350 THE MOTIVATION OF BEHAVIOR

cular system activity for problems of emotion and motivation.
According to Hebb, activation produced by the diffuse bombard-
ment of the cortex by the arousal system is ". . . synonymous with
a general drive state, and the conception of drive therefore assumes
anatomical and physiological identity" (p. 249). Schlosberg (1954)
has also stressed the emotional and motivational aspects of arousal
activity, but he has not written expressly of its possible relations
to Hull's general-purpose drive. These writers, and others (e.g.,
Malmo, 1958), have also observed that performance decrements
characteristic of "overly motivated" subjects might reasonably be
attributed to excessive bombardment of the cortex by the ascending
reticular activating system. On this view, performance is related
to motivation ( activation ) level by an inverted U-shaped function,
performance being optimally efficient when amount of activation
is neither too weak (the subject is drowsy or relaxed) nor too
strong (the subject is angry, fearful, or enraged).

Support for the view that impulses in the ascending reticular
system function in a relatively nonspecific fashion is provided
primarily by observed changes in electrocortical potentials. But
scattered data of a more molar nature are also consistent with
this view. For example, it has long been known that responses of
flight, rage, and generally heightened "emotionality" can be pro-
duced by intracranial stimulation; and it is responses of this sort
that have often been seen as reflecting heightened drive. Moreover,
according to Ingram, Knott, and Chiles (1953) and Lindsley
(1957), electrical stimulation of the brain-stem reticular system
has alerting or arousing effects upon overt behavior as well as upon
the pattern of the electroencephalogram. Perhaps the most inter-
esting data of this kind are those reported by Fuster (1958).
According to this investigator, stimulation of the reticular forma-
tion of monkeys (at the mesencephalic level) while they are en-
gaged in the performance of visual discrimination tasks increases
their speed of reacting, improves their discriminatory accuracy,
and lowers their tachistoscopic recognition thresholds. Precisely
what the mechanism of this effect may be is not clear, but it seems
to be congruent with the idea of a nonspecific drive. In several
studies, however (e.g., Chiles, 1954; and Knott, Ingram, and Cor-
rell, 1960), lever pressing in cats tends to be arrested rather than

MISCELLANEOUS MOTIVATIONAL PROBLEMS 351

augmented by electrical stimulation in the region of the hypo-
thalamus.

Several investigators (Delgado, Roberts, and Miller, 1954; Cohen,
Brown, and Brown, 1957; Roberts, 1958; Bursten and Delgado,
1958) have shown that intracranial stimulation may function in
much the same way as peripheral shock. To take a single example,
Delgado and his associates have noted that electrical stimulation
of certain thalamic, mesencephalic, and rhinencephalic structures
elicits fearlike responses in cats. The resultant emotionality can
be conditioned to external stimuli; it can function to motivate
the learning of instrumental acts; and it can serve as punishment
in training hungry cats to avoid food. Although part of this effect
may have been due to cerebral pain, similar results were not ob-
tained from the stimulation of sensorimotor areas. Apparently,
therefore, centrally introduced stimuli, when applied to certain
structures and under certain learning conditions, play a role in
the determination of behavior comparable to that of peripheral
shocks or other drive-arousing unconditioned stimuli.

The above discussion has stressed the nonspecific consequences
of central stimulation, but quite specific responses are also fre-
quently evoked. According to Samuels (1959), relatively specific
effects of central stimulation are more probably reflections of
activity in the thalamic reticular system than of action in the brain-
stem reticular system. As an example of such effects. Miller (1958)
reports the elicitation of well-integrated eating and drinking re-
sponses in satiated rats as the result of central (hypothalamic)
stimulation. This specificity of action is not of primary concern
here, but it may constitute the physiological basis for such behavior-
directing effects as are commonly ascribed to drive-variable stimuli
{SnS).

Considered as a receiving station for incoming impulses, the
reticular system is thought to be affected primarily by signals
diverted via collaterals from the primary sensory pathways and,
in addition, by impulses descending from the cortex. This means,
as Lindsley (1957), Samuels (1959), and others have indicated,
that the activity level of the multineuronal, multisynaptic reticular
formation can be modified by both exteroceptive and interoceptive
stimuli and even by "ideational" impulses presumed to originate

352 THE MOTIVATION OF BEHAVIOR

in the higher cortical centers. From this the conclusion may be
drawn that, in so far as the alerting and arousing effects of reticular
system discharges can be equated to nonspecific drive, then level
of drive is under the control of external and/or internal stimuli.
The implications of this suggestion are widespread and of con-
siderable importance, since, if this is indeed the case, plausible
mechanisms are provided for (1) increasing drive by presenting
the CS for an aversive response or by presenting an incentive
object, (2) reducing drive by presenting a secondarily reinforcing
stimulus, or (3) changing drive in almost any manner by verbal
instructions or self-induced ideations. As we have previously ob-
served in our discussion of the concepts of drive and need (Chap-
ter 3), the possibility that drive strength can be changed by
exteroceptive or proprioceptive stimuli may clarify some of the
problems posed by the view that drive reduction is essential for
learning.

It also seems probable that activity in the reticular system can
control the sensitivity of peripheral receptors and, at a more
central level, can integrate and coordinate receptor-generated
information. The verification of these functions by further research
may provide us with mechanisms by which drive can affect sensa-
tion, perception, memory, and a variety of other processes.

Reinforcing Properties of Intracerebral Stimuli. Although the
behavioral effects of electrical stimulation of the brain have been
studied in both acute and chronic preparations for many years.
Olds and Milner (1954) were apparently the first to show that
such stimuli can function as reinforcers for instrumental acts. This
finding has provided the impetus both for considerable further
research and for theoretical reappraisals of the mechanisms of
reinforcement.

In the Olds-Milner demonstrations, electrodes were perma-
nently implanted in the brains of rats in such a manner as not to
interfere with normal daily activities. A pair of thin, flexible wires
could be attached, as desired, to the electrodes where they emerged
from the skull. These wires were suspended from above and hence
did little to hamper the rat's movements in a Skinner box or
through simple mazes. This arrangement made it possible for the
experimenters to administer intracerebral electric stimuli of any

MISCELLANEOUS MOTIVATIONAL PROBLEMS 353

strength or duration, at any moment, and in any relation to the
overt actions of the rat.

The most interesting finding reported by Olds and Milner was
that after appropriate conditions of pretraining, some subjects,
particularly those having electrode tips in the limbic system,
would press a Skinner-box lever repeatedly and for long periods
of time, even though each lever-pressing response was accompanied
by an intracranial shock. In some cases the shocks lasted until
the lever was released, but usually their duration was fixed at 0.5
or I.O second. Under these conditions, with no rewards other than
electrical stimuli being provided, rats would continue to press the
lever for periods of several hours and at rates as high as approxi-
mately 30 responses per minute. Unquestionably, "masochistic"
behavior of this kind is puzzling, since electric shock has been so
widely used as an aversive or punishing stimulus. Nevertheless,
the phenomenon is now well authenticated, having been repeatedly
demonstrated in rats (Olds, 1956ci, 1956/?; Reynolds, 1958; Bower
and Miller, 1958), in cats (Sidman, Brady, Boren, Conrad, and
Schulman, 1955; Roberts, 1958), and in monkeys (Bursten and
Delgado, 1958). There is also fairly general agreement that stimu-
lation of the limbic system, and especially in the ventral region
of the anterior hypothalamus, is more likely to yield positive
reinforcement effects than the stimulation of other regions. The
effect is not confined, however, to a single sharply circumscribed
area, and since rather high intensities of shock are often used,
structures other than those adjacent to the electrode tips may be
involved.

When consideration is given to the question of why these sub-
cortical stimuli function as reinforcers, no entirely convincing
answers are forthcoming. One suggestion is that pleasurable sensa-
tions or experiences are aroused when known reinforcing areas are
excited. On this view, certain portions of the brain are seen as
"pleasure centers," since organisms do perform responses accom-
panied by the stimulation of these regions. Other areas, the stim-
ulation of which seems to evoke avoidance and withdrawal rather
than approach, would qualify as "pain centers." This conception
is attractive to advocates of hedonistic theories of reinforcement
and is supported by introspective reports of "pleasurable" sensa-

354 THE MOTIVATION OF BEHAVIOR

tions by human subjects who, while conscious, have been stimu-
lated in or near the septal region (Heath, 1954). The view is also
consistent with the notion that some of the reinforcing areas of
the brain may exert a quieting or inhibiting effect on behavior.
Thus Brady and Nauta (1953) have shown that extirpation of the
septal area leads to an increase in activity, and that perhaps, there-
fore, the normal function of that area is generally inhibitory. As
Olds (1955) remarks, ongoing activity usually stops abruptly when
the septal region is stimulated electrically.

Weighing against this pleasure-center view are the findings of
Roberts (1958), Bower and Miller (1958), and Brown and Cohen
(1959). All these investigators have shown that stimulation of a
given region, even with intensity held constant, may be either
reinforcing or inhibiting depending upon the training procedures
used and upon such factors as stimulation duration and the testing
environment. Roberts, for instance, reports that cats in a Y maze
will learn to enter one arm of the maze where shock to the pos-
terior hypothalamus is turned on but will also learn to enter a
second arm to turn off the shock. Apparently the onset of excita-
tion is reinforcing, since the animals repeatedly enter the maze
arm where shock is turned on; but continuation of the stimulus
seemingly becomes aversive, because the cats also learn to enter
the only arm of the maze where the shock is always terminated.
Of special interest is Roberts's finding that increases in stimulus
intensity affect the approach and escape responses differentially.
With relatively weak shocks the response leading to shock onset
is readily learned, whereas the turning-off response is not. But with
high-voltage stimulation marked improvement is shown in the
learning of the turning-off response while the turning-on reaction
is performed more slowly and with clear indications of conflict.

Qualitative reports of the behavior of animals stimulated in
reinforcing areas also weigh against any simple pleasure-center
interpretation. Thus Olds (1955) observes that, following pro-
longed periods of reinforcing self-stimulation in the septal area,
rats become vicious and are dangerous to handle. This seems to
contradict the notion that septal stimulation has pleasurable or
soothing consequences. Moreover, the typical behavior pattern
exhibited by both rats and cats during periods of intermittent self-

MISCELLANEOUS MOTIVATIONAL PROBLEMS 355

Stimulation is often described as "agitated searching" or even as
"flight," which scarcely suggests relaxation and contentment. If
the self-stimulation of the sort investigated in these studies is
pleasurable, it is the frenetic, compulsive pleasure of the masochist
or the neurotic. Apparently neither rats nor cats, when stimulus
duration is made to coincide with bar-depression time, hold the
bar down for protracted periods so that the self-stimulation can
be prolonged. Were this the case, rate of lever pressing should be
extremely low, since the subject should have little tendency to
release the lever and thereby terminate the "pleasant" experience.
High rates of lever pressing are usually obtained with high voltages,
indicating that the subjects, though they repeatedly approach and
depress the lever, must be releasing it quickly and thus terminating
the shock. Why should the lever be released if this act eliminates
a genuinely pleasurable stimulus? The answer to this question is no
clearer than that to the question of why the rat depresses the
lever in the first place. In any event, although rats certainly develop
a strong tendency to approach and press the lever, they also learn
to release it after experiencing a relatively brief shock.

It is also quite probable that the success of self-stimulation
demonstrations depends in considerable degree upon the voltage
levels used during early training trials, upon the rapidity with
which the voltage is increased over trials, and upon other aspects
of the training and testing procedures. Olds (1955), for instance,
reports that his rats were manually placed upon the bar to receive
their first shock, that they were replaced thereon whenever they
were not responding regularly, and that the progressive upward
shifting of voltages had to be carefully executed in order to achieve
maximum response rates. As yet no one seems to have set the
shock at a high level initially and then left the animals alone to
see whether they would then learn to press the lever to stimulate
themselves. Judging from Roberts's results, if strong shock were
used from the outset, it would probably not be reinforcing, though
the same intensity of stimulation, if arrived at through a series of
carefully graded steps, might so function. Such an experiment has
apparently not even been done with mild shock, and, to the writer's
knowledge, no learning curves have been published in which data
from every shock trial (including the first) have been presented.

356 THE MOTIVATION OF BEHAVIOR

The importance of training and testing procedures is further high-
hghted by the work of Bursten and Delgado (1958), who failed
in a preliminary attempt to teach monkeys to press a lever for
intracranial shock reward but succeeded in demonstrating the rein-
forcing effects of shock by the use of a shuttle box.

There is also merit in an interpretation proposed by Cohen,
Brown, and Brown (1957). These investigators have observed
that stimulation of intracranial structures such as the hypothala-
mus, though it may produce motivational increments, does not
immediately and consistently evoke directed escape responses. The
fearlike or ragelike emotionality cannot be referred by the animal
to any specific objects or regions in his environment. Central
shock is thus unlike grid shock, since the latter, because of the
direction-giving peripheral pain cues, regularly elicits movements
of leaping, hopping, or running that are directed away from the
source of irritation. On this view, when animals are reinforced
with central stimuli, as in the Olds-Milner study, they do not
avoid the lever, because the central shock provides the animal
with no information to indicate that the lever or any other par-
ticular feature of the testing environment is the source of the
stimulus. Or to phrase it differently, mild central shock does not
elicit lever-escape responses, and until these are evoked and are
reinforced, the animals have no reason to refrain from approaching
the lever. The fact that animals learn to press the lever is not, of
course, explained by this interpretation. It is conceivable that fac-
tors such as those operating in studies of experimental masochism
(see below) may be of significance in connection with this prob-
lem.

In the early flush of enthusiasm generated by studies of intra-
cranial reinforcing stimuli, the fact that quite similar effects have
been reported for externally administered shock and other strong
stimuli seems to have escaped notice. Masserman (1946), for
instance, in his studies of "experimental masochism," trained cats
to obtain food by operating a lever. Blasts of air of gradually
increasing intensity were then administered each time the lever
was depressed. At the end of training, the cats were not only not
disturbed by airblasts intense enough to evoke violent avoidance
in nontrained animals, but they even operated the switch lever in

MISCELLANEOUS MOTIVATIONAL PROBLEMS 357

the absence of food reward, the airblast having apparently acquired
the properties of a substitute reward. Likewise, Farber (1948) and
Miller and Davis (1943) have demonstrated that after rats have
been trained to run down an alley for food, mild electric grid
shock can be introduced in the middle of the alley without disrupt-
ing the approach behavior. The latter investigators also report that
if shock strength is then increased gradually over trials, it can even-
tually reach a very high level without causing the animals to cease
running, and that, in fact, their speed of running is increased.
But if the voltage is raised precipitously, abrupt cessation of run-
ning is the result. A comparable phenomenon, observed a number
of years ago by the writer, has been described as the "vicious-
circle" phenomenon by Mowrer (1950). Rats were trained initially
to run down a short straight alley to escape shock by reaching a
nonelectrified section at the end. After this behavior was firmly
established, the electrical connections to the first three feet of the
grid were removed to make it "safe," while the last three feet were
left electrified. Some rats, when placed into the safe section at the
start, continued to run from the no-shock starting section and
across the shock section to the safe goal region. With some
animals this behavior persisted for literally hundreds of trials
without the slightest sign of abatement. Mowrer has suggested
that conditioned fear provides the motivation for running out of
the safe starting section, that strength of fear is maintained by
the shock encountered on the way to the safe goal, and that fear
reduction, operating through the law of effect, continues to rein-
force running. If to this is added the consideration that shock
reduction also provides a powerful and continuing reinforcement
for running, the plausibility of the theory is enhanced. Be that
as it may, when rats exhibit this "vicious-circle" behavior they are
responding quite like the animals in the intracranial self-stimula-
tion experiments. It is even possible that an interpretation such
as Mowrer's, which seems to provide a satisfactory explanation for
one of these phenomena, can be extended to encompass the other.
We have already noted in discussing the so-called manipulatory
and exploratory drives that rats will learn to press a lever when
each press is followed by the onset of a light (Kish, 1955). This
phenomenon, sometimes described by the phrase "sensory-change

358 THE MOTIVATION OF BEHAVIOR

reinforcement," has been amply confirmed by Marx, Henderson,
and Roberts (1955), by Kling, Horowitz, and Delhagen (1956),
and by Barnes and Kish (1957). It is of interest here because of
the ways in which it parallels the phenomena of reinforcement
by central stimulation. For one thing, if the light is made too
bright, its onset ceases to be reinforcing and becomes aversive
(Marx et al.). Clearly, this is similar to Roberts's (1958) observa-
tion that when a central stimulus is intense, "turning-off" responses
are easily learned but "turning-on" responses are not. Moreover,
lever pressing both for central stimulation ( Brady, Boren, Conrad,
and Sidman, 1957; Olds, 1958) and for light onset (Forgays and
Levin, 1959) is increased by food deprivation. And both types
of stimulation, if administered under the proper conditions, can
apparently reinforce more complicated responses than lever press-
ing, for example, maze behavior (Olds, 1955) and discrimination
learning (Forgays and Levin, 1959). In the light of these similari-
ties, it is conceivable that both types of experiments involve the
operation of basically similar mechanisms of reinforcement.

The implications of findings like these for the drive-reduction
hypothesis of reinforcement have been pointed out elsewhere
(Brown, 1955). The fact that organisms will perform responses
leading to, or accompanied by, increased intensity of stimulation
is not especially damaging to this hypothesis. In the first place,
there are many instances in which enhanced physical stimulation
need not, and probably should not, be described as leading to or
generating increments in magnitude of drive. In each particular
case one must consider not only the amount of physical change
but also the adaptation level of the subject's sense organs, the
nature of his ongoing responses, his homeostatic balances, and his
previous experiences with the stimulus. If, when evaluated in terms
of these and other factors, a stimulus does not qualify as a source
of drive, then the continued appearance of responses that are
followed by that stimulus is not conclusive evidence against the
drive-reduction position. In the second place, we must insist once
more that the concepts of drive and need are not identical or at
least do not have to be identical. A bodily need may function as a
source of drive, but the class of drive-reducing operations contains

MISCELLANEOUS MOTIVATIONAL PROBLEMS 359

more members than the class of need-reducing conditions. The
body's need for water can be eliminated only by water or by fluids
containing water, but the drive arising from that need can perhaps
be partially reduced, at least temporarily, by such activities as
turning one's attention to other matters, by chewing a stick of gum,
by swallowing repeatedly, and so on. Inasmuch as all these activities
involve an increase in stimulation, it again becomes apparent that
drive level can probably be modified by either external or internal
stimuli. It would be premature, therefore, to insist that drive is not
reduced in self-stimulation demonstrations simply because no
biological need is apparently satisfied. But it would be equally
unwise to hold stubbornly to a drive-reduction view until one can
point with some assurance to the source of the drive that is being
reduced by self-stimulation and to the mechanisms by which this
reduction might be achieved.

As a third and final point, one must bear in mind that neither
self-administered intracranial stimuli nor exteroceptive stimuli can
serve to motivate the behavior of approaching and pressing a bar,
if, as is usually the case, they are not turned on until after the
to-be-learned response has been initiated. In this respect, the self-
stimulation studies are like the Butler-Harlow experiments on
visual rewards with monkeys. The motivation underlying prestimu-
lation behavior may be a kind of conditioned anxiety, or a positive
expectancy (rg-Sg) of anticipated pleasure, or perhaps simply
the animal's normally present general drive. The first of these
possibilities is consistent with Mowrer's interpretation of the
"vicious-circle" phenomenon, and the second with the view that
the onset of some stimuli may be "pleasurable" and that classically
conditioned fractional responses anticipatory of such pleasure might
serve as sources of drive. As to the third possibility, little need be
said save that normal, healthy, awake organisms may always be
operating under moderate drive even when no extreme biological
imbalances exist. Taken in conjunction with the manifold stimuli
provided by the usual environment, this persisting drive may
suffice to activate a wide variety of behaviors, whose probabilities
of appearance are subsequently modified by reinforcements of one
sort or another.

360 THE MOTIVATION OF BEHAVIOR

Summary

It has been our puq)ose in this final chapter to expand our treat-
ment of motivation by considering several topics that could not
conveniently have been included in earlier chapters.

One of these topics involves the problem of whether nonnoxious
stimuli that evoke responses of approaching, exploring, and manip-
ulating also tend to function as motivational variables. Spence and
Hull hold that such stimuli may indeed generate increments to
drive, provided they have been closely associated in the past with
reinforcing events. Thus responses composing appetitive consum-
matory sequences are thought to become conditioned to any stimuli
present when those responses occur, and fractional components of
the sequences can also be evoked by similar stimuli even prior to
the time the goal is reached. These fractional anticipatory goal
reactions {rgs) are said to affect behavior both associatively, because
of the characteristic stimuli {SgS) they are assumed to provide,
and motivationally, by adding an increment to drive.

Since organisms may approach, inspect, and manipulate unfamil-
iar as well as familiar objects, the possibility must be entertained
that novel stimuli also function motivationally. With unfamiliar
stimuli, however, interpretations involving acquired, anticipatory
responses are clearly inappropriate. As a result, various writers
(Montgomery, Harlow, Berlyne, and others) have urged that the
postulation of special-purpose exploratory, curiosity, or manipula-
tory drives is desirable. These drives, however, are drives to perform
only those specific kinds of acts that are evoked by the presentation
of novel stimulus objects. Thus they involve both behavior-direct-
ing and motivating functions; and in the role of systematic con-
cepts, these behavior-specific drives suffer from the fact that they
are seldom defined in terms of observable events other than the
particular actions they are alleged to produce.

Representative experiments dealing with exploratory, investiga-
tive, and manipulatory behavior indicate that rats, as well as higher
organisms, do approach and investigate some, though by no means
all, strange or novel objects. However, when such behavior is not
evoked, or when actual withdrawal occurs, a fear drive is invoked
in lieu of an exploratory drive. Nevertheless, numerous investiga-

MISCELLANEOUS MOTIVATIONAL PROBLEMS 361

tions can be cited which support the conclusion that organisms
often behave so as to increase their contacts with unfamihar stimuh
or locales and to reduce their contacts with familiar objects or
regions. And although exploratory and investigative behavior weak-
ens rather quickly as novelty diminishes, manipulatory behavior in
monkeys is remarkably persistent. In addition, some students of
behavior have concluded that an opportunity to investigate a
strange environment serves as a reinforcing agent for the learning
of new responses.

Our analyses of these experiments and their interpretations
indicate that in most instances it is difficult to reject the hypothesis
that motivational factors of an aversive nature have been present
in addition to, or instead of, the alleged investigative or manipula-
tory drives. Continued contact with, or confinement in, one seg-
ment of a maze may be an aversive condition that is alleviated
by movement to other segments. It is also uncertain, in the case
of experiments in which monkeys are reinforced by the act of
looking out of an opaque-walled box, whether appeal must be
made to the drive-arousing attributes of the external cues. At the
very least, the drive underlying the performance of responses
occurring prior to seeing cannot be ascribed to the effects of stimuli
encountered only after seeing is permitted. However, mild anxiety
due to isolation and confinement, or auditory stimuli, might func-
tion in such situations as motivators of preseeing behavior. It
remains to be demonstrated, therefore, that special-purpose drives
must be invoked to explain the variety of actions usually described
as exploratory and/or manipulatory.

Learned emotional reactions are said by many to have significant
motivational properties. The question of how such responses are
learned, therefore, merits consideration in any theory of motivation.
Some theorists (e.g., Guthrie and Hull) find no reasons for sup-
posing that the learning of autonomically governed reactions is
due to factors or processes different from those involved in the
learning of skeletal (instrumental) responses. Others, however,
maintain that different principles apply to the learning of the two
classes of reactions. For example, Mowrer regards stimulus-response
contiguity as both the necessary and the sufficient condition for
the learning of fears, but drive reduction is seen as a further neces-

362 THE MOTIVATION OF BEHAVIOR

sary condition for the acquisition of instrumental acts. This view
contrasts with Hull's supposition that both drive reduction and
S-R contiguity are necessary for the learning of all reactions.

Our analysis of these views and of experiments designed to
support them reveals that a lack of precision in the structure of
all extant theories makes crucial tests difficult. Experiments that
have been reported appear to be unusually susceptible to interpre-
tations other than those presented by the investigators and are
marred by misconstructions of opposing theories and by the con-
founding effects of uncontrolled variables.

In the third major section of the chapter, evidence from phys-
iological studies of central nervous system activity is reviewed with
respect to its implications for a concept of general drive. Here the
most significant finding appears to be that stimulation in the
region of the brain-stem reticular formation can change electrical
potentials at many parts of the cortex from a pattern characteristic
of sleep to one associated with alert wakefulness. The motivational
implications of this widespread bombardment of the cortex by the
ascending reticular activating system have been widely noted, and
Hebb has specifically identified this activity with a general-drive
state. Apparently, these alerting functions can be modified by a
wide variety of influences, including internal and external stimuli,
and even the so-called ideational and attitudinal factors presum-
ably mediated largely by cortical processes. It thus appears that
drive level, to the extent that it may be identified with arousal
activity of the ascending reticular system, can either be raised or
lowered by a multiplicity of events. The potential significance of
this possibility for such problems as those of secondary reinforce-
ment, perceptual defense or vigilance, and the drive-reduction
theory of reinforcement remains to be explored.

The final section of the chapter deals with studies of the rein-
forcing and inhibiting effects of centrally administered brain
stimuli. The fact that animals can learn to perform acts such as
lever pressing when the only consequence of the act is a sub-
cortical electric shock is the empirical phenomenon of central
interest. The basic finding has been repeatedly confirmed in rats
and cats and, with some success, in monkeys. Some investigators
have been led to the conclusion that reinforcing regions of the

MISCELLANEOUS MOTIVATIONAL PROBLEMS 363

brain might be construed as "pleasure centers," whereas other
areas, the stimulation of which leads to inhibition of activity,
might be termed "pain centers." Support for this view has been
provided by introspective reports from human subjects who have
been stimulated while undergoing brain surgery under local anes-
thesia. In a variety of studies, however, it has been found that an
animal can be taught either to perform an act to turn the brain
stimulus on, or to perform a different act to turn it off. Since in
these demonstrations the electrode placement and shock intensity
is the same, any simple concept of pleasure or pain centers becomes
relatively untenable. Apparently brain stimuli can have either
rewarding or punishing effects, depending upon the kinds of
behavior that are elicited by the test situation and upon temporal
and other relations between the stimuli and the behavior.

REFERENCES

iiiiiniiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiH^

Adolph, E. F. (1941): The internal environment and behavior: III.
Water content. Amer. J. Psychiat., 97, 1365-1373.

Adolph, E. F. (1943): Physiological regulations. New York: The
Ronald Press Company.

Amsel, A. (1950a): The combination of a primary appetitional need
with primary and secondary emotionally derived needs. /. exp.
Psychol, 40, 1-14.

Amsel, A. (1950b): The effect upon level of consummatory response
of the addition of anxiety to a motivational complex. /. exp.
Psychol, 40, 709-715.

Amsel, A. (1951): A three-factor theory of inhibition: an addition to
Hull's two-factor theory. Amer. Psychologist, 6, 487. (Abstract.)

Amsel, A. (1958) : The role of frustrative nonreward in noncontinuous
reward situations. Psychol. Bull, SS, 102-119.

Amsel, A., and W, Hancock (1957): Motivational properties of
frustration: III. Relation of frustration effect to antedating goal
factors. /. exp. Psychol, 53, 126-131.

Amsel, A., and I. Maltzman (1950): The effect upon generalized
drive strength of emotionality as inferred from the level of con-
summatory response. /. exp. Psychol, 40, 563-569.

Amsel, A., and J. Roussel (1952): Motivational properties of frustra-
tion: I. Effect on a running response of the addition of frustration
to the motivational complex. /. exp. Psychol, 43, 363-368.

Anand, B. K., and J. R. Brobeck (1951): Hypothalamic control of
food intake in rats and cats. Yale J. Biol, and Med., 24, 123-140.

Anderson, E. E. (1941): The externalization of drive: III. Maze
learning by non-rewarded and by satiated rats. J. genet. Psychol,
59, 397-426.

Armus, H. L. (1954) : The effect of percentage of reinforcement and
distribution of trials on resistance to extinction of a conditioned
fear response. Unpublished doctoral dissertation. State University
of Iowa.

365

366 THE MOTIVATION OF BEHAVIOR

Aronson, E. (1956): The need for achievement as measured by
graphic expression. Unpublished master's thesis, Wesleyan Uni-
versity. Cited in McClelland, 1956.

Ashley, W. R., R. S. Harper, and D. L. Runyon (1951): The per-
ceived size of coins in normal and hypnotically induced economic
states. Amer. J. Psychol, 64, 564-572.

Atkinson, J. W. (1950) : Studies in projective measurement of achieve-
ment motivation. University of Michigan. Abstract in University
Microfilms, Vol. X, N. 4; Publication No. 1945. Cited in McClel-
land, Atkinson, Clark, and Lowell, 1953.

Atkinson, J. W. (1954): Explorations using imaginative thought to
assess the strength of human motives. In M. R. Jones (ed.),
Nebraska symposium on motivation. Lincoln, Nebr.: University
of Nebraska Press.

Atkinson, J. W., and D. C. McClelland (1948): The projective ex-
pression of needs. II. The effect of different intensities of the
hunger drive on thematic apperception. J. exp. Psychol., 38,
643-658.

Bahrick, H. P. (1953) : Sensory preconditioning under two degrees of
deprivation. /. comp. physiol. Psychol., 46, 39-42.

Bakan, D. (1953): Learning and the scientific enterprise. Psychol.
Rev., 60, 45-49.

Baker, R. A. (1955): The effects of repeated deprivation experience
on feeding behavior. J. comp. physiol. Psychol., 48, 37-42.

Barnes, G. W., and G. B. Kish (1957) : Reinforcing properties of the
termination of intense auditory stimulation. /. comp. physiol.
Psychol, 50, 40-43.

Barry, H., Ill (1958): Effects of strength of drive on learning and
extinction. /. exp. Psychol, 55, 473-481.

Bass, Bettina (1958) : The effect of drive variation within and between
subjects on conditioning performance. Unpublished doctoral dis-
sertation, State University of Iowa.

Beams, H. L. (1954): Affectivity as a factor in the apparent size of
pictured food objects. /. exp. Psychol, 47, 197-200.

Bechtoldt, H. P., and J. S. McDonough (1958): The effects of
motivating instructions upon visual identification thresholds. (Un-
published. Personal communication.)

Bergmann, G. (1944): An empiricist's system of the sciences. Sci.
Monthly, 59, 140-148.

Berkun, M. M., M. L. Kessen, and N. E. Miller, (1952): Hunger-
reducing effects of food by stomach fistula versus food by mouth
measured by a consummatory response. /. comp. physiol. Psychol,
45, 550-554.

Berlyne, D. E. (1955) : The arousal and satiation of perceptual curios-
ity in the rat. /. comp. physiol Psychol, 48, 238-246,

REFERENCES 367

Berlyne, D. E. (1958): The present status of research on exploratory
and related behavior. /. indiv. Psychol., 14, 121-126.

Bills, A. G. (1927): The influence of muscular tension on the effi-
ciency of mental work. Amer. J. Psychol., 38, 227-251.

Bindra, D. (1959) : Stimulus change, reactions to novelty, and response
decrement. Psychol. Rev., 66, 96-103.

Bindra, D., A. L. Paterson, and J. Strzelecki (1955) : On the relation
between anxiety and conditioning. Canad. J. Psychol., 9, 1-6.

Birch, D., E. Burnstein, and R. A. Clark (1958) : Response strength as
a function of food deprivation under a controlled maintenance
schedule. J. comp. physiol. Psychol., SI, 350-354.

Bitterman, M. E., and W. H. Holtzman (1952): Conditioning and
extinction of the galvanic skin response as a function of anxiety.
/. abnorm. soc. Psychol., 47, 615-623.

Bitterman, M. E., and C. W. KnifEn (1953): Manifest anxiety and
"perceptual defense." /. abnorm. soc. Psychol., 48, 248-252.

Bitterman, M. E., P. Reed, and J. Krauskopf (1952): The effect of
the duration of the unconditioned stimulus upon conditioning
and extinction. Amer. J. Psychol., 65, 256-262.

Bolles, R. C. (1958) : Comments on Professor Estes' paper. In M. R.
Jones (ed.), Nebraska symposium on motivation. Lincoln, Nebr.:
University of Nebraska Press.

Bower, G. H., and N. E. Miller (1958): Rewarding and punishing
effects from stimulating the same place in the rat's brain. /. comp.
physiol. Psychol, 51, 669-674.

Brady, J. V., }. J. Boren, D. Conrad, and M. Sidman (1957): The
effect of food and water deprivation upon intracranial self-stimu-
lation. /. comp. physiol. Psychol., 50, 134-137.

Brady, J. V., and W. J. H. Nauta (1953) : Subcortical mechanisms in
emotional behavior: affective changes following septal forebrain
lesions in the albino rat. /. comp. physiol. Psychol., 46, 339-346.

Brady, J. V., W. C. Stebbins, and R. Galambos (1953) : The effect of
audiogenic convulsions on a conditioned emotional response.
J. comp. physiol. Psychol., 46, 363-367.

Brandauer, C. M. (1953):A confirmation of Webb's data concerning
the action of irrelevant drives. /. exp. Psychol., 45, 150-152.

Braunstein, E. P. (1923): Zur Lehre von den kurzdauernden Lich-
treizen der Netzhaut. Zsc/i. /. Sinnesphysiol., 55, 185-229. Cited
by S. Hecht (1934): The nature of the photoreceptor process.
In C. Murchison (ed.). Handbook of general experimental psy-
chology. Worcester, Mass.: Clark University Press, 1934.

Bricker, P. D., and A. Chapanis (1953): Do incorrectly perceived
tachistoscopic stimuli convey some information? Psychol. Rev.,
60, 181-188.

Broadhurst, P. L. (1957): Emotionality and the Yerkes-Dodson law.
/. exp. Psychol, 54, 345-352.

368 THE MOTIVATION OF BEHAVIOR

Brobeck, J. R. (1955): Neural regulation of food intake. In R. W.
Miner (ed.), The regulation of hunger and appetite. Ann. N. Y.
Acad. Sci., 63, Art. 1, New York: N.Y. Acad. Sci.

Brown, D. R., and J. Adams (1954) : Word frequency and the meas-
urement of value areas. /. abnorm. soc. Psychol., 49, 427-430.

Brown, G. W., and B. D. Cohen (1959): Avoidance and approach
learning motivated by stimulation of identical hypothalamic
nuclei. Amer. J. Physiol, 197, 153-157.

Brown, J. S. (1948): Gradients of approach and avoidance responses
and their relation to level of motivation. /. comp. physiol. Psychol.,
41, 450-465.

Brown, J. S. (1953(z) : Problems presented by the concept of acquired
drives. In Current theory and research in motivation: a sym-
posium. Lincoln, Nebr.: University of Nebraska Press.

Brown, J. S. (1953b): Comments on Professor Harlow's paper. In
Current theory and research in motivation: a symposium. Lincoln,
Nebr.: University of Nebraska Press.

Brown, J. S. (1955) : Pleasure-seeking behavior and the drive-reduction
hypothesis. Psychol. Rev., 62, 169-179.

Brown, J. S. (1957): Principles of intrapersonal conflict. Conflict
Resolution, 1, 135-154.

Brown, J. S., and I. E. Farber (1951): Emotions conceptualized as
intervening variables — with suggestions toward a theory of frus-
tration. Psychol. Bull, 48, 465-495.

Brown, J. S., and A. Jacobs (1949) : The role of fear in the motivation
and acquisition of responses. J. exp. Psychol, 39, 747-759.

Brown, J. S., H. I. Kalish, and I. E. Farber (1951) : Conditioned fear
as revealed by magnitude of startle response to an auditory
stimulus. J. exp. Psychol, 41, 317-328.

Brown, J. S., J. W. Meryman, and F. N. Marzocco (1956): Sound-
induced startle response as a function of time since shock. /.
comp. physiol Psychol, 49, 190-194.

Brown, W. L., and G. Gentry (1948) : The effects of intra-maze delay.
II. Various intervals of delay. /. comp. physiol. Psychol, 41, 403-
407.

Brozek, J., H. Guetzkow, and M. V. Baldwin (1951): A quantitative
study of perception and association in experimental semistarva-
tion. /. Pers., 19, 245-264.

Bruner, J. S., and C. C. Goodman (1947) : Value and need as organ-
izing factors in perception. /. abnorm. soc. Psychol, 42, 33-44.

Bruner, J. S., and L. Postman (1947): Emotional selectivity in per-
ception and reaction. /. Pers., 16, 69-77.

Bruner, J. S., and L. Postman (1948) : Symbolic value as an organizing
factor in perception. /. soc. Psychol, 27, 203-208.

Bruner, J. S., and J. S. Rodrigues (1953): Some determinants of
apparent size. /. abnorm. soc. Psychol, 48, 17-24.

REFERENCES 369

Bugelski, R., and N. E. Miller (1938): A spatial gradient in the
strength of avoidance responses. /. exp. Psychol., 23, 494-505.

Burros, R. H. (1949): Quantitative studies in secondary drive. Un-
published doctoral dissertation, Yale University.

Bursten, B., and J. M. R. Delgado (1958): Positive reinforcement
induced by intracerebral stimulation in the monkey. /. comp.
physiol. Psychol., 5 J, 6-10.

Butler, R. A. (1953): Discrimination learning by rhesus monkeys to
visual-exploration motivation. J. comp. physiol. Psychol., 46,
95-98.

Butler, R. A. (1957): The eflFect of deprivation of visual incentives
on visual exploration motivation in monkeys. /. comp. physiol.
Psychol, SO, 177-179.

Butler, R. A. (1958): Exploratory and related behavior: a new trend
in animal research. J. indiv. Psychol., 14, 111-120.

Butler, R. A., and H. F. Harlow (1954): Persistence of visual ex-
ploration in monkeys. /. comp. physiol. Psychol., 47, 258-
263.

Calvin, J. S., E. A. Bicknell, and D. S. Sperling (1953): Establish-
ment of a conditioned drive based on the hunger drive. /. comp.
physiol. Psychol, 46, 173-175.

Campbell, B. A., and D. Kraeling (1953): Response strength as a
function of drive level and amount of drive reduction. /. exp.
Psychol, 4S, 97-101.

Campbell, B. A., and D. Kraeling (1954): Response strength as a
function of drive level during training and extinction. /. comp.
physiol Psychol, 47, 101-103.

Campbell, N. R. (1921) : What is science? London: Methuen & Co.,
Ltd.

Cannon, W. B. (1929): Bodily changes in pain, hunger, fear, and
rage, 2d ed. New York: Appleton-Century-Crofts, Inc.

Carlton, P. L. (1955): The effect of time of food deprivation on
selective learning. Unpublished doctoral dissertation, State Uni-
versity of Iowa.

Carter, L. F., and K. Schooler (1949) : Value, need, and other factors
in perception. Psychol. Rev., S6, 200-207.

Castaneda, A., B. R. McCandless, and D. S. Palermo (1956): The
children's form of the manifest anxiety scale. Child Develpm.,
17, 317-326.

Castaneda, A., D. S. Palermo, and B. R. McCandless (1956): Com-
plex learning and performance as a function of anxiety in children
and task difficulty. Child Develpm., 27, 327-332.

Champion, R. A. (1954): Drive-strength and competing response
tendencies. Unpublished master's thesis, State University of Iowa.

Chapanis, A., W. R. Garner, and C. T. Morgan (1949)': Applied
experimental psychology. New York: John Wiley & Sons, Inc.

370 THE MOTIVATION OF BEHAVIOR

Chapman, R. M., and N. Levy (1957) : Hunger drive and reinforcing
effect of novel stimuli. /. comp. physiol. Psychol., SO, 233-238.

Child, I. L., and I. K. Waterhouse (1953): Frustration and the
quality of performance: II. A theoretical statement. Psychol.
Rev., 60, 127-139.

Chiles, W. D. (1954): Performance during stimulation of the dien-
cephalic activating system. /. comp. physiol. Psychol., 47, 412-
415.

Chodorkoff, B. (1956): Anxiety, threat, and defensive reactions. /.
gen. Psychol, 54, 191-196.

Clark, R. A. (1952): The projective measurement of experimentally
induced levels of sexual motivation. J. exp. Psychol., 44, 391-399.

Clark, R. A., and D. C. McClelland (1950) : A factor analytic integra-
tion of imaginative, performance, and case study measures of
the need for achievement. Unpublished paper, cited in McClel-
land, Atkinson, Clark, and Lowell, 1953.

Clark, R. E. (1958) : Size estimation limens as a function of level of
motivation. (Unpublished. Personal communication.)

Cohen, B. D., G. W. Brown, and M. L. Brown (1957): Avoidance
learning motivated by hypothalamic stimulation. /. exp. Psychol.,
53, 228-233.

Cohen, M. R., and E. Nagel (1934): An introduction to logic and
scientific method. New York: Harcourt, Brace and Company, Inc.

Conger, J. J. (1951): The effects of alcohol on conflict behavior in
the albino rat. Quart. J. Stud. Alcohol, 12, 1-29.

Cooper, J. B. (1938): The effect upon performance of introduction
and removal of a delay within the maze. J. comp. Psychol., 25,
457-462.

Cotton, J. W. (1953): Running time as a function of amount of
food deprivation. /. exp. Psychol., 46, 188-198.

Courts, F. A. (1942): Relations between muscular tension and per-
formance. Psychol. Bull, 39, 347-367.

Cowles, J. T. (1937) : Food- tokens as incentives for learning by chim-
panzees. Comp. Psychol. Monogr. 14, No. 5 (Whole No. 71).

Crespi, L. P. (1944): Amount of reinforcement and level of per-
formance. Psychol. Rev., 51, 341-357.

Davenport, J. W. (1956): Choice behavior as a function of drive
strength and rate of learning. Unpublished doctoral dissertation,
State University of Iowa.

Davis, J. D. (1958): The reinforcing effect of weak-light onset as a
function of amount of food deprivation. J. comp. physiol. Psychol,
51, 496-498.

Davis, R. C. (1953): Physical psychology. Psychol Rev., 60, 7-14.

Davis, R. H. (1957): The effect of drive reversal on latency, ampli-
tude, and activity level. /. exp. Psychol, 53, 310-315.

REFERENCES 371

Davitz, J. R. (1955): Reinforcement of fear at the beginning and at
the end of shock. J. comp. physiol. Psychol., 48, 152-155.

Deese, J., and J. A. Carpenter (1951) : Drive level and reinforcement.
/. exp. Psychol, 42, 236-238.

Deese, J., R. S. Lazarus, and }. Keenan (1953): Anxiety, anxiety re-
duction, and stress in learning. /. exp. Psychol., 46, 55-60.

Delgado, J. M. R., W. W. Roberts, and N. E. Miller (1954) : Learn-
ing motivated by electrical stimulation of the brain. Amer. J.
Physiol, 179, 587-593.

Dember, W. N., and H. Fowler (1958): Spontaneous alteration
behavior. Psychol Bull, SS, 412-428.

Dinsmoor, J. A. (1952): The effect of hunger on discriminated re-
sponding. /. abnorm. soc. Psychol, 47, 67-72.

Dodson, J. D. (1917): Relative values of reward and punishment in
habit formation. Psychobiology, 1, 231-276.

Dollard, J., and N. E. Miller (1950): Personality and psychotherapy.
New York: McGraw-Hill Book Company, Inc.

Dollard, J., L. W. Doob, N. E. Miller, O. H. Mowrer, and R. R. Sears
(1939): Frustration and aggression. New Haven, Conn.: Yale
University Press.

Duffy, E. (1934) : Emotion: an example of the need for reorientation
in psychology. Psychol. Rev., 41, 184-198.

Duffy, E. (1951) : The concept of energy mobilization. Psychol Rev.,
SB, 30-40.

Duffy, E. (1957): The psychological significance of "arousal" or
"activation." Psychol Rev., 64, 265-275.

Dukes, W. F., and W. Bevan, Jr. (1952) : Size estimation and mone-
tary value: a correlation. J. Psychol, 34, 43-53.

Eisman, E. (1956): An investigation of the parameters defining drive
(D). J. exp. Psychol, 52, 85-89.

Eisman, E., A. Asimow, and I. Maltzman (1956) : Habit strength as a
function of drive in a brightness discrimination problem. /. exp.
Psychol, S2, 58-64.

Eriksen, C. W. (1951): Some implications for TAT interpretation
arising from need and perception experiments. /. Pers., 19,
282-288.

Eriksen, C. W. (1954): The case for perceptual defense. Psychol
Rev., 61, 175-182.

Estes, W. K. (1950) : Toward a statistical theory of learning. Psychol.
Rev., 57, 94^107.

Estes, W. K. (1958): Stimulus-response theory of drive. In M. R.
Jones (ed.), Nebraska symposium on motivation. Lincoln, Nebr.:
University of Nebraska Press.

Estes, W. K., and B. F. Skinner (1941) : Some quantitative properties
of anxiety. /. exp. Psychol, 29, 390-400.

372 THE MOTIVATION OF BEHAVIOR

Farber, I. E. (1948) : Response fixation under anxiety and non-anxiety
conditions. J. exp. Psychol., 38, 111-131.

Farber, I. E. (1954d) : Anxiety as a drive state. In M. R. Jones (ed.),
Nebraska symposium on motivation. Lincoln, Nebr.: University
of Nebraska Press.

Farber, I. E. (1945b): Comments on Professor Atkinson's paper.
In M. R. Jones (ed.), Nebraska symposium on motivation. Lin-
coln, Nebr.: University of Nebraska Press.

Farber, I. E. (1955): The role of motivation in verbal learning and
performance. Psychol. Bull., 52, 311-327.

Farber, 1. E., and K. W. Spence (1953) : Complex learning and condi-
tioning as a function of anxiety. /. exp. Psychol., 4S, 120-125.

Finch, G. (1942): Chimpanzee frustration responses. Psychosom.
Med., 4, 233-251.

Forgays, D. G., and H. Levin (1959): Discrimination and reversal
learning as a function of change of sensory stimulation. /. comp.
physiol. Psychol, SI, 191-194.

Franks, CM. (1957) : Effect of food, drink, and tobacco deprivation
on the conditioning of the eyeblink response. J. exp. Psychol., S3,
117-120.

Fredenburg, N. C. (1956): Response strength as a function of alley
length and time of deprivation. Unpublished master's thesis, State
University of Iowa.

Freeman, J. T. (1954) : Set or perceptual defense? /. exp. Psychol, 48,
283-288.

French, E. G. (1956) : Development of a measure of complex motiva-
tion. Research Report TN-56-48, AF Personnel and Training
Research Center, Lackland AF base, Texas.

French, J. R. P., Jr. (1944) : Organized and unorganized groups under
fear and frustration. In R. R. Sears (ed.), Authority and frustra-
tion: Studies in topological and vector psychology. HI. University
Iowa Stud. Child Welfare, 20, 231-308.

Freud, S. (1905, First German ed.): Three essays on sexuality. In
Standard Edition. Vol. VII. London: Hogarth Press, Ltd., 1953.

Freud, S. (1936) : The problem of anxiety. New York: W. W. Norton
& Company, Inc.

Fuster, J. M. (1958) : Effects of stimulation of brain stem on tachisto-
scopic perception. Science, 127, 150.

Gardiner, H. M., R. C. Metcalf, and J. G. Beebe-Center (1937):
Feeling and emotion. New York: American Book Company,

Gerall, A. A., P. B. Sampson, and G. L Boslov (1957): Classical
conditioning of human pupillary dilation. /. exp. Psychol, S4,
467-474.

Ghent, L. (1957) : Some effects of deprivation on eating and drinking
behavior. /. comp. physiol. Psychol, SO, 172-176.

Gilbert, R. W. (1936): The effect of noninformative shock upon

REFERENCES 373

maze learning and retention with human subjects. J. exp. Psychol.,
19, 456-466.

Gilbert, R. W. (1937): A further study of the effect of noninforma-
tive shock upon learning. J. exp. Psychol., 20, 396-407.

Gilchrist, J. C., and L. S. Nesberg (1952): Need and perceptual
change in need-related objects. /. exp. Psychol., 44, 369-376.

Gilhousen, H. C. (1938) : Temporal relations in anticipatory reactions
of the white rat in a maze. /. comp. Psychol., 26, 163-175.

Glanzer, M. (1953): The role of stimulus satiation in spontaneous
alternation. /. exp. Psychol, 45, 387-393.

Glanzer, M. (1958): Curiosity, exploratory drive, and stimulus satia-
tion. Psychol. Bull, 55, 302-315.

Grice, G. R., and J. D. Davis (1957): Effect of irrelevant thirst
motivation on a response learned with food reward. J. exp.
Psychol, 53, 347-352.

Grice, G. R., and E. Saltz (1950): The generalization of an instru-
mental response to stimuli varying in the size dimension. J. exp.
Psychol, 40, 702-708.

Grossman, M. I. (1955): Integration of current views on the regula-
tion of hunger and appetite. In R. W. Miner (ed.). The regula-
tion of hunger and appetite. Ann. N. Y. Acad. Sci., 63, Art. 1,
New York: N. Y. Acad. Sci.

Guthrie, E. R. (1935) : The psychology of learning. New York: Harper
& Brothers.

Guthrie, E. R., and G. P. Horton (1946) : Cats in a puzzle box. New
York: Holt, Rinehart and Winston, Inc.

Gwinn, G. T. (1951): Resistance to extinction of learned fear-drives.
/. exp. Psychol, 42, 6-12.

Haigh, G. V., and D. W. Fiske (1952): Corroboration of personal
values as selective factors in perception. /. abnorm. soc. Psychol,
47, 394-398.

Hall, C. S., and G. Lindzey (1957): Theories of personality. New
York: John Wiley & Sons, Inc.

Hall, J. F., and P. V. Hanford (1954): Activity as a function of a
restricted feeding schedule. /. comp. physiol Psychol, 47, 362-
363.

Hamilton, G. V. (1916): A study of perseverance reactions in pri-
mates and rodents. Behav. Monogr., 3, No. 2 (Whole No. 13).

Haner, C. F., and P. A. Brown (1955) : Clarification of the instigation
to action concept in the frustration-aggression hypothesis. /.
abnorm. soc. Psychol, 51, 204-206.

Harlow, H. F. (1953): Motivation as a factor in the acquisition of
new responses. In Current theory and research in motivation: a
symposium. Lincoln, Nebr.: University of Nebraska Press.

Heath, R. G. (1954) : Behavioral changes following destructive lesions
in the subcortical structures of the forebrain in cats. In R. G.

374 THE MOTIVATION OF BEHAVIOR

Heath (ed.), Studies in schizophrenia; a multidisciplinary ap-
proach to mid-brain relationships. Cambridge, Mass.: Harvard
University Press.

Heathers, G. L., and P. Arakehan (1941): The relation between
strength of drive and rate of extinction of a bar-pressing reaction
in the rat. /. gen. Psychol, 24, 243-258.

Hebb, D. O. (1949) : The organization of behavior. New York: John
Wiley & Sons, Inc.

Hebb, D. O. (1955): Drives and the c.n.s. (conceptual nervous sys-
tem). Psychol. Rev., 62, 243-254.

Hedlund, J. L. (1953): Construction and evaluation of an objective
test of achievement imagery. Unpublished doctoral dissertation.
State University of Iowa.

Hilden, A. H. (1937): An action current study of the conditioned
hand withdrawal. In L. E. Travis (ed.), Studies in Clinical Psy-
chology. Psychol. Monogr., 49, No. 1 (Whole No. 217).

Hilgard, E. R., and D. G. Marquis (1935): Acquisition, extinction,
and retention of conditioned lid responses to light in dogs. /.
comp. Psychol, 19, 29-58.

Hilgard, E. R., and D. G. Marquis (1940): Conditioning and learn-
ing. New York: Appleton-Century-Crofts, Inc.

Hilgard, E. R., L. V. Jones, and S. J. Kaplan (1951): Conditioned
discrimination as related to anxiety. J. exp. Psychol, 42, 94-99.

Hillman, B., W. S. Hunter, and G. A. Kimble (1953) : The effect of
drive level on the maze performance of the white rat. /. comp.
physiol Psychol, 46, 87-89.

Holder, E. (1951): Response strength as a function of changes in
the length of the delay interval. Unpublished master's thesis.
New Mexico State College. Cited in Holder, Marx, Holder and
Collier, 1957.

Holder, W. B., M. H. Marx, E. E. Holder, and G. Collier (1957):
Response strength as a function of delay of reward in a runway.
/. exp. Psychol, 53, 316-323.

Hollenberg, E., and M. Sperry (1951): Some antecedents of aggres-
sion and effects of frustration in doll play. Personality: topical
symposia, 1, 32-43.

Holton, R. B. (1956) : Variables affecting the change in instrumental
response magnitude after reward cessation. Unpublished doctoral
dissertation, State University of Iowa.

Horenstein, B. R. (1951): Performance of conditioned responses as a
function of strength of hunger drive. /. comp. physiol Psychol,
44, 210-224.

Hovland, C. I. (1936) : "Inhibition of reinforcement" and phenomena
of experimental extinction. Proc. Nat. Acad. Sci., Washington, 22,
430-433.

REFERENCES 375

Hovland, C. I., and A. H. Riesen (1940) : Magnitude of galvanic and
vasomotor response as a function of stimulus intensity. /. gen.
Psychol, 23, 103-121.

Howes, D. H., and R. L. Solomon (1950): A note on McGinnies'
"Emotionality and perceptual defense." Psychol. Rev., S7, 229-
234.

Hughes, J. B., II, J. L. Sprague, and A. W. Bendig (1954): Anxiety
level, response alternation, and performance in serial learning. /.
Psychol, 38, 421-426.

Hull, C. L. (1931): Goal attraction and directing ideas conceived as
habit phenomena. Psychol Rev., 38, 487-506.

Hull, C. L. (1937) : Mind, mechanism, and adaptive behavior. Psychol
Rev., 44, 1-32.

Hull, C. L. (1943) : Principles of behavior. New York: Appleton-Cen-
tury-Crofts, Inc.

Hull, C. L. (1951): Essentials of behavior. New Haven, Conn.: Yale
University Press.

Hull, C. L. (1952) : A behavior system. New Haven, Conn.: Yale Uni-
versity Press.

Hunt, H. F., and J. V. Brady (1951): Some effects of electroconvul-
sive shock on a conditioned emotional response ("anxiety").
/. comp. physiol Psychol, 44, 88-98.

Hunt, H. F., and L. S. Otis (1953): Conditioned and unconditioned
emotional defecation in the rat. /. comp. physiol. Psychol, 46,
378-382.

Hurley, J. R. (1955) : The Iowa picture interpretation test: a multiple-
choice variation of the TAT. /. consult. Psychol, 19, 372-376.

Hurley, J. R. (1957): Achievement imagery and motivational instruc-
tions as determinants of verbal learning. /. Pers., 25, 274-282.

Ingram, W. R., J. R. Knott, and W. D. Chiles (1953): Behavioral
and electrocortical effects of diencephalic stimulation in un-
anesthetized, unrestrained cats. XJX Internat. Physiol. Congr.,
487-488.
anowitz, H. D., and M. I. Grossman (1949): Some factors affecting
the food intake of normal dogs and dogs with esophagostomy
and gastric fistula. Amer. J. Physiol, 159, 143-148.
asper, H. H., (ed.) (1958) : Reticular formation of the brain. Boston:

Little, Brown & Company,
enkin, N. (1957): Affective processes in perception. Psychol. Bull,

54, 100-127.
erome, E. A., J. A. Moody, T. J. Connor, and M. B. Fernandez
(1957): Learning in a multiple-door situation under various
drive states. /. comp. physiol. Psychol, 50, 588-591.
ohanson, A. M. (1922) : The influence of incentive and punishment
upon reaction time. Arch. Psychol, 8, No. 54.

376 THE MOTIVATION OF BEHAVIOR

Johnston, R. A. (1955): The effects of achievement imagery on

maze-learning performance. J. Pers., 24, 145-152.
Johnston, R. A. (1957): A methodological analysis of several revised

forms of the Iowa Picture Interpretation Test. J. Pers., 25, 283-

293.
Jones, M. C. (1924): The elimination of children's fears. J. exp.

Psychol, 7, 382-390.
Kabrick, R. P., and I. E. Farber (1952): Effect of electric shock on

speed of reaction. Amer. Psychologist, 7, 586. (Abstract.)
Kagan, J., and M. Berkun (1954): The reward value of running

activity. J. comp. physiol. Psychol., 47, 108.
Kalish, H. I. (1954): Strength of fear as a function of the number

of acquisition and extinction trials. /. exp. Psychol., 47, 1-9.
Kaplan, M. (1952): The effects of noxious stimulus intensity and

duration during intermittent reinforcement of escape behavior.

J. comp. physiol. Psychol, 45, 538-549.
Kendler, H. H. (1945): Drive interaction: I. Learning as a function

of the simultaneous presence of the hunger and thirst drives.

/. exp. Psychol, 35, 96-109.
Ketchel, R. G. (1955): Performance in instrumental learning as a

function of shock intensity. Unpublished master's thesis, State

University of Iowa.
Kimble, G. A. (1951) : Behavior strength as a function of the intensitv

of the hunger drive. /. exp. Psychol, 41, 341-348.
Kish, G. B. (1955): Learning when the onset of illumination is used

as reinforcing stimulus. /. comp. physiol. Psychol, 48, 261-

264.
Klein, G. S., H. J. Schlesinger, and D. E. Meister (1951): The effect

of personal values on perception: an experimental critique. Psy-
chol Rev., 58, 96-112.
Kling, J. W., L. Horowitz, and J. E. Delhagen (1956): Light as a

positive reinforcer for rat responding. Psychol. Rep., 2, 537-340.
Knott, J. R., W. R. Ingram, and R. E. Correll (I960): Some effects

of subcortical stimulation on the bar press response. A. M. A.

Arch. Neurol, 2, 476-484.
Koch, S., and W. J. Daniel (1945): The effect of satiation on the

behavior mediated by a habit of maximum strength. J. exp.

Psychol, 35, 167-187.
Lacy, O. W., N. Lewinger, and J. F. Adamson (1953): Foreknowl-
edge as a factor affecting perceptual defense and alertness. /. exp.

Psychol, 45, 169-174.
Lambert, W. W., R. L. Solomon, and P. D. Watson (1949): Rein-
forcement and extinction as factors in size estimation. /. exp.

Psychol, 39, 637-641.
Lawrence, D. LI., and W. A. Mason (1955): Intake and weight

REFERENCES 377

adjustments in rats to changes in feeding schedule. J. comp.
physiol Psychol, 48, 43-46.

Lawson, R., and M. H. Marx (1958) : Frustration: theory and experi-
ment. Genet. Psychol. Monogr., SI, 393-464.

Lazarus, R. S., J. Deese, and S. F. Osier (1952): The effects of psy-
chological stress upon performance. Psychol. Bull, 49, 293-317.

Lazarus, R. S., C. W. Eriksen, and C. P. Fonda (1951): Personality
dynamics and auditory perceptual recognition. J. Pers., 19, 471-
482.

Lazarus, R. S., and R. A. McCleary (1951) : Autonomic discrimination
without awareness: a study of subception. Psychol. Rev., S8,
113-122.

Lazarus, R. S., H. Yousem, and D. Arenberg (1953): Hunger and
perception. J. Pers., 21, 312-328.

Levine, R., L Chein, and G. Murphy (1942): The relation of the
intensity of a need to the amount of perceptual distortion: a
preliminary report. /. Psychol., 13, 283-293.

Lewin, K. (1931): Environmental forces in child behavior and devel-
opment. In C. Murchison (ed.), A handbook of child psychology.
Worcester, Mass.: Clark University Press.

Lewin, K. (1938): The conceptual representation and the measure-
ment of psychological forces. Contr. Psychol. Theor., I, No. 4.

Libby, A. (1951): Two variables in the acquisition of depressant
properties by a stimulus. J. exp. Psychol., 42, 100-107.

Lindsley, D. B. (1951): Emotion. In S. S. Stevens (ed.), Handbook
of experimental psychology. New York: John Wiley & Sons, Inc.

Lindsley, D. B. (1957): Psychophysiology and motivation. In M. R.
Jones (ed.), Nebraska symposium on motivation. Lincoln, Nebr.:
University of Nebraska Press.

Lindsley, D. B., J. Bowden, and H. W. Magoun (1949) : Effect upon
EEG of acute injury to the brain stem activating system. EEC
din. Neurophysiol, I, 475-486.

Lindzev, G. (ed.) (1959) : Assessment of human motives. New York:
Holt, Rinehart and Winston.

Loess, H. B. (1952) : The effect of variation of motivational level and
changes in motivational level on performance in learning. Unpub-
lished doctoral dissertation. State University of Iowa.

Logan, F. A. (1952) : The role of delay of reinforcement in determin-
ing reaction potential. J. exp. Psychol, 43, 393-399.

Logan, F. A., D. L. Olmstead, B. S. Rosner, R. D. Schwartz, and
C. M. Stevens (1955): Behavior theory and social science. New
Haven, Conn.: Yale University Press.

Lowell, E. L. (1952cz): The effect of conflict on motivation. Unpub-
lished doctoral dissertation. Harvard University. Cited in Whit-
ing and Child, 1953.

378 THE MOTIVATION OF BEHAVIOR

Lowell, E. L. (1952b) : The effect of need for achievement on learning
and speed of performance. J. Psychol., 33, 31-40.

Ludvigson, W. H. (1958) : Visual identification thresholds as a func-
tion of attitude. (Unpublished. Personal communication.)

Lysak, W. (1954): The effects of punishment upon syllable recogni-
tion thresholds. /. exp. Psychol, 47, 343-350.

Lysak, W., and J. C. Gilchrist (1955): Value, equivocality, and goal
availability as determinants of size judgments. /. Pers., 23,
500-501.

McAllister, W. R. (1953) : The effect on eyehd conditioning of shift-
ing the CS-US interval. J. exp. Psychol., 4S, 423-428.

McCandless, B. R., and A. Castaneda (1956): Anxiety in children,
school achievement, and intelligence. Child Develpm., 27, 379-
382.

McClelland, D. C. (1951): Personality. New York: William Sloane
Associates.

McClelland, D. C. (1956) : The measurement of human needs. Final
Report, Contract N7 onr 463, NR 171-363, OfEce of Naval Re-
search, Aug. 31, 1956.

McClelland, D. C, and J. W. Atkinson (1948): The projective ex-
pression of needs: I. The effects of different intensities of the
hunger drive on perception. /. Psychol., 2S, 205-222.

McClelland, D. C, J. W. Atkinson, R. A. Clark, and E. L. Lowell
(1953) : The achievement motive. New York: Appleton-Century-
Crofts, Inc.

McClelland, D. C, and A. M. Liberman (1949): The effect of need
for achievement on recognition of need-related words. /. Pers., 18,
236-251.

MacCorquodale, K., and P. E. Meehl (1953): Preliminary sugges-
tions as to a formalization of expectancy theory. Psychol. Rev.,
60, 55-63.

McDougall, W. (1917): An introduction to social psychology, 12th
ed. Boston: Luce.

McGinnies, E. (1949) : Emotionality and perceptual defense. Psychol.
Rev., S6, 244-251.

McGinnies, E. (1950): Discussion of Howes' and Solomon's note on
"Emotionality and perceptual defense." Psychol. Rev., 57, 235-
240.

McGinnies, E., and H. Sherman (1952) : Generalization of perceptual
defense. /. abnorm. soc. Psychol., 47, 81-85.

Madsen, K. B. (1959): Theories of motivation. Copenhagen: Munks-
gaard.

Malmo, R. B. (1957): Anxiety and behavioral arousal. Psychol. Rev.,
64, 276-287.

Malmo, R. B. (1958) : Measurement of drive: an unsolved problem in

REFERENCES 379

psychology. In M. R. Jones (ed.), Nebraska symposium on mo-
tivation. Lincoln, Nebr.: University of Nebraska Press.

Marx, M. H. (1956): Some relations between frustration and drive.
In M. R. Jones (ed.), Nebraska s}>mposium on motivation. Lin-
coln, Nebr.: University of Nebraska Press.

Marx, M. H., R. L. Henderson, and C. L. Roberts (1955): Positive
reinforcement of the bar-pressing response by a light stimulus
following dark operant pretests with no aftereffect. /. comp.
physiol. Psychol, 48, 73-76.

Marzocco, F. N. (1951): Frustration effect as a function of drive
level, habit strength, and distribution of trials during extinction.
Unpublished doctoral dissertation. State University of Iowa.

Masserman, J. H. (1946): Principles of dynamic psychiatry. Philadel-
phia: W. B. Saunders Company.

Mausner, B., and A. Siegel (1950) : The effect of variation in "value"
on perceptual thresholds. /. abnorm. soc. Psychol., 45, 760-763.

Melton, A. W. (1941): Learning. In W. S. Monroe (ed.). Encyclo-
pedia of educational research. New York: The Macmillan Com-
pany.

Meryman, J. J. (1952) : Magnitude of startle response as a function of
hunger and fear. Unpublished master's thesis. State University of
Iowa.

Meryman, J. J. (1953) : The magnitude of an unconditioned GSR as
a function of fear conditioned at a long CS-UCS interval. Unpub-
lished doctoral dissertation. State University of Iowa.

Meyer, D. R. (1951): Food deprivation and discrimination reversal
learning by monkeys. /. exp. Psychol., 41, 10-16.

Meyer, D. R. (1953): On the interaction of simultaneous responses.
Psychol. Bull, SO, 204-220.

Meyer, D. R., and M. E. Noble (1958): Summation of manifest
anxiety and muscular tension, /. exp. Psychol, SS, 599-602.

Miller, N. E. (1944): Experimental studies of conflict. In J. McV.
Hunt (ed.). Personality and the behavior disorders. Vol. I. New
York: The Ronald Press Company.

Miller, N. E. (1948d): Studies of fear as an acquirable drive: I. Fear
as motivation and fear-reduction as reinforcement in the learning
of new responses. J. exp. Psychol, 38, 89-101.

Miller, N. E. (1948b) : Tlieory and experiment relating psychoanabtic
displacement to stimulus-response generalization. /. abnorm.
soc. Psychol, 43, 155-178.

Miller, N. E. (1951): Learnable drives and rewards. In S. S. Stevens
(ed.). Handbook of experimental psychology. New York: John
Wiley & Sons, Inc.

Miller, N. E. (1958): Central stimulation and other new approaches
to motivation and reward. Amer. Psychologist, 13, 100-108.

380 THE MOTIVATION OF BEHAVIOR

Miller, N. E. (1959) : Liberalization of basic S-R concepts: Extensions
to conflict behavior, motivation, and social learning. In S. Koch
(ed.). Psychology: a study of a science. Vol. 2. General sys-
tematic formulations, learning, and special processes. New York:
McGraw-Hill Book Companv, Inc.

Miller, N. E., and M. Davis (1943): A theoretical and experimental
analysis of conflict behavior: IV. The influence of the positions of
reward and punishment in the response sequence. (Unpublished
paper. Cited in Miller, 1944.)

Miller, N. E., and J. Dollard (1941): Social learning and imitation.
New Haven, Conn.: Yale University Press.

Miller, N. E., and D. H. Lawrence (1950): Studies of fear as an ac-
quirable drive: III. Effect of strength of electric shock as a pri-
mary drive and of number of trials with the primarv drive on
the strength of fear. (Unpublished paper. Cited in Miller, 19 5 1.)

Miller, N. E., R. I. Sampliner, and P. Woodrow (1957): Thirst-
reducing effects of water by stomach fistula vs. water by mouth
measured by both a consummatory and an instrumental response.
/. comp. physiol. Psychol., SO, 1-5.

Miller, N. E., and S. S. Stevenson. (1936): Agitated behavior of rats
during experimental extinction and a curve of spontaneous re-
covery. /. comp. Psychol., 21, 205-231.

Miller, W. C, and J. E. Greene (1954) : Generalization of an avoid-
ance response to varving intensities of sound. J. comp. physiol.
Psychol, 47, 136-139.

Miner, R. W. (ed.) (1955): The regulation of hunger and appetite.
Ann. N.Y. Acad. Sci., 63, Art. 1., New York: N.Y. Acad. Sci.

Moeller, G. (1954): The CS-UCS interval in GSR conditioning. /.
exp. Psychol, 48, 162-166.

Montgomery, K. C. (1952): A test of two explanations of spon-
taneous alternation. /. comp. physiol. Psychol, 4S, 287-293.

Montgomery, K. C. (1953): Exploratory behavior as a function of
"similarity" of stimulus situations. /. comp. physiol. Psychol,
46, 129-133.

Montgomery, K. C. (1954) : The role of the exploratory drive in learn-
ing. J. comp. physiol. Psychol, 47, 60-64.

Montgomery, K. C. (1955): The relation between fear induced by
novel stimulation and exploratory behavior. /. comp. physiol.
Psychol, 48, 254-260.

Montgomery, K. C, and M. Segall (1955): Discrimination learning
based upon the exploratory drive. /. comp. physiol. Psychol, 48,
225-228.

Morgan, C. T. (1943): Physiological psychology. New York: Mc-
Graw-Hill Book Company, Inc.

REFERENCES 381

Morgan, C. T. (1957): Physiological mechanisms of motivation. In
M. R. Jones (ed.), Nebraska symposium on motivation. Lincoln,
Nebr.: University of Nebraska Press.

Morgan, C. T., and E. Stellar (1950): Physiological psychology, 2d
ed. New York: McGraw-Hill Book Company, Inc.

Moruzzi, G., and H. W. Magoun (1949) : Brain stem reticular forma-
tion and activation of the EEG. EEG din. Neurophysiol., 1,
455-473.

Mowrer, O. H. (1939): A stimulus-response analysis of anxiety and
its role as a reinforcing agent. Psychol. Rev., 46, 553-565.

Mowrer, O. H. (1947) : On the dual nature of learning — a re-interpre-
tation of "conditioning" and "problem-solving." Harvard Educ.
Rev., 17, 102-148.

Mowrer, O. H. (1950): Learning theory and personality dynamics.
New York: The Ronald Press Company.

Mowrer, O. H. (1951): Two-factor learning theory: summary and
comment. Psychol. Rev., 58, 350-354.

Mowrer, O. H., and E. G. Aiken (1954): Contiguity vs. drive-reduc-
tion in conditioned fear: temporal variations in conditioned and
unconditioned stimulus. Amer. J. Psychol., 67, 26-38.

Mowrer, O. H., and R. R. Lamoreaux (1942): Avoidance condition-
ing and signal duration — a study of secondary motivation and re-
ward. Psychol. Monogr., S4, No. 5 (Whole No. 247).

Mowrer, O. H., and L. N. Solomon (1954): Contiguity vs. drive-
reduction in conditioned fear: the proximity and abruptness of
drive-reduction. Amer. J. Psychol., 67, 15-25.

Muenzinger, K. F. (1934): Motivation in learning. II. The function
of electric shock for right and wrong responses in human subjects.
/. exp. Psychol, 17, 439-448.

Murfin, F. L. (1954) : The relationship of fear to the CS-UCS interval
in acquisition and the CS-duration in extinction. Unpublished
doctoral dissertation. State University of Iowa.

Murphy, D. B. (1953): Recognition and recall as a function of fre-
quency, shock, and individual differences variables. Unpublished
doctoral dissertation. State University of Iowa.

Murray, H. A. (1938) : Explorations in personality. New York: Oxford
University Press.

Myers, A. K., and N. E. Miller (1954) : Failure to find a learned drive
based on hunger; evidence for learning motivated by "explora-
tion." J. comp. physiol. Psychol, 47, 428-436.

Myers, T. I. (1952): An experimental investigation of the effect of
hunger drive upon the brightness discrimination learning of
the rat. Unpublished doctoral dissertation. State University
of Iowa.

382 THE MOTIVATION OF BEHAVIOR

Nagaty, M. O. (1951) : The effect of reinforcement on closely follow-
ing S-R connections: I. The effect of a backward conditioning
procedure on the extinction of conditioned avoidance. /. exp.
Psychol, 42, 239-246.

Neel, A. F. (1954) : Conflict, recognition time and defensive behavior.
Amer. Psychologist, 9, 437-438. (Abstract.)

Nissen, H. W. (1954) : The nature of the drive as innate determinant
of behavioral organization. In M. R. Jones (ed.), Nebraska sym-
posium on motivation. Lincoln, Nebr.: University of Nebraska
Press.

Olds, J. (1955) : Physiological mechanisms of reward. In M. R. Jones
(ed.), Nebraska symposium on motivation. Lincoln, Nebr.:
University of Nebraska Press.

Olds, J. (1956d): A preliminary mapping of electrical reinforcing ef-
fects in the rat brain. J. comp. physiol. Psychol., 49, 281-285.

Olds, J. ( 1956b) : Runway and maze behavior controlled by basomedial
forebrain stimulation in the rat. /. comp. physiol. Psychol., 49,
507-512.

Olds, J. (1958): Effects of hunger and male sex hormone on self-
stimulation of the brain. J. comp. physiol. Psychol., SI, 320-324.

Olds, J., and P. Milner (1954): Positive reinforcement produced by
electrical stimulation of septal area and other regions of rat brain.
/. comp. physiol. Psychol, 47, 419-427.

Osgood, C. E. (1953) : Method and theory in experimental psychology.
New York: Oxford University Press.

Palermo, D. S., A. Castaneda, and B. R. McCandless (1956): The
relationship of anxiety in children to performance in a complex
learning task. Child Develpm., 27, 333-337.

Passey, G. E. (1948): The influence of intensity of unconditioned
stimulus upon acquisition of a conditioned response. /. exp. Psy-
chol, 38, 420-428.

Pavlov, I. P. (1927) : Conditioned reflexes. London: Oxford University
Press.

Perin, C. T. (1942): Behavior potentially as a joint function of the
amount of training and the degree of hunger at the time of ex-
tinction. /. exp. Psychol, 30, 93-113.

Postman, L. (1947): The history and present status of the law of
effect. Psychol Bull, 44, 489-563.

Postman, L. (1953d): The experimental analysis of motivational fac-
tors in perception. In Current theory and research in motivation:
a symposium. Lincoln, Nebr.: University of Nebraska Press.

Postman, L. (1953jb) : Comments on papers by Professors Brown and
Harlow. In Current theory and research in motivation: a sym-
posium. Lincoln, Nebr.: University of Nebraska Press.

Postman, L., W. C. Bronson, and G. L. Cropper (1953): Is there

REFERENCES 383

a mechanism of perceptual defense? /. abnorm. soc. Psychol., 48,
215-224.

Postman, L., J. S. Bruner, and E. McGinnies (1948) : Personal values
as selective factors in perception. /. abnorm. soc. Psychol., 43,
142-154.

Postman, L., and R. S. Crutchfield (1952): The interaction of need,
set and stimulus-structure in a cognitive task. Amer. J. Psychol.,
65, 196-217.

Powloski, R. F. (1953): The effects of combining hunger and thirst
motives in a discrimination habit. /. comp. physiol. Psychol., 46,
434-437.

Proshansky, H., and G. Murphy (1942): The effects of reward and
punishment on perception. J. Psychol., 13, 293-305.

Pustell, T. E. (1957): The experimental induction of perceptual
vigilance and defense. J. Pers., 25, 425-438.

Ramond, C. K. (1954): Performance in selective learning as a func-
tion of hunger. /. exp. Psychol., 48, 265-270.

Reece, M. M. (1954) : The effect of shock on recognition thresholds.
/. abnorm. soc. Psychol., 49, 165-172.

Reese, T. W. (1943) : The application of the theory of physical meas-
urement to the measurement of psychological magnitudes, with
three experimental examples. Psychol. Monogr., 55, No. 3
(Whole No. 251).

Reynolds, R. W. (1958): The relationship between stimulation volt-
age and rate of hypothalamic self-stimulation in the rat. J. comp.
physiol. Psychol, 51, 193-198.

Richter, C. P. (1933): The effect of early gonadectomy on the gross
bodily activity of rats. Endocrinology, 17, 445-450.

Roberts, W. W. (1958): Both rewarding and punishing effects from
stimulation of posterior hypothalamus of cat with same electrode
at same intensity. /. comp. physiol. Psychol., 51, 400-407.

Rohrer, J. H. (1949) : A motivational state resulting from non-reward.
/. comp. physiol. Psychol, 42, 476-485.

Rosenbaum, G. (1953) : Stimulus generalization as a function of level
of experimentally induced anxiety. /. exp. Psychol, 45, 35-43.

Rosenthal, B. G. (1951): Attitude towards money, need, and meth-
ods of presentation as determinants of perception of coins from
six to ten years of age. Amer. Psychologist, 6, 317. (Abstract.)

Rossi, G. F., and A. Zanchetti (1957): The brain stem reticular for-
mation: anatomy and physiology. Arch, ital Biol, 95, 199-435.

Roussel, J. (1952): Frustration effect as a function of repeated non-
reinforcements and as a function of the consistency of reinforce-
ment prior to the introduction of nonreinforcement. Unpub-
lished master's thesis, Tulane University. Cited in Amsel, 1958.

Saltzman, I., and S. Koch (1948): The effect of low intensities of

384 THE MOTIVATION OF BEHAVIOR

hunger on the behavior mediated by a habit of maximum
strength. /. exp. Psychol, 38, 347-370.'

Samuels, I. (1959): Reticular mechanisms and behavior. Psvchol.
Bull, 56, 1-25.

Sanford, R. N. (1936): The effects of abstinence from food upon
imaginal processes: a preliminary experiment. /. Psychol, 2, 129-
136.

Sanford, R. N. (1937): The effects of abstinence from food upon
imaginal processes: a further experiment. /. Psychol, 3, 145-
159.

Schlosberg, H. (1937) : The relationship between success and the laws
of conditioning. Psychol Rev., 44, 379-394.

Schlosberg, H. (1954): Three dimensions of emotion. Psychol Rev.,
61, 81-88.

Seward, }. P. (1951) : Experimental evidence for the motivating func-
tion of reward. Psychol Bull, 48, 130-149.

Seward, J. P. (1956) : A neurological approach to motivation. In M. R.
Jones (ed.), Nebraska symposium on motivation. Lincoln, Nebr.:
University of Nebraska Press.

Sheffield, V. F. (1950): Resistance to extinction as a function of the
distribution of extinction trials. /. exp. Psychol, 40, 305-313.

Sheriffs, A. C. (1948): The "intuition questionnaire": a new projec-
tive test. /. abnorm. soc. Psychol, 43, 326-331 .

Shirley, M. (1928): Studies in activity. II. Activity rhythms; age and
activity; activity after rest. /. comp. Psychol, 8, 159-186.

Sidman, M., |. V. Brady, }. J. Boren, D. G. Conrad, and A. Schul-
man (1955): Reward schedules and behavior maintained by in-
tracranial self-stimulation. Science, 122, 830-831.

Siegel, P. S. (1943) : Drive shift, a conceptual and experimental analy-
sis. /. comp. Psychol, 35, 139-148.

Siegel, P. S. (1947) : The relationship between voluntary water intake,
body weight loss, and number of hours of water privation in the
rat. /. comp. physiol Psychol, 40, 231-238.

Siegel, P. S., and J. J. Brantley (1951): The relationship of emotion-
ality to the consummatory response of eating. /. exp. Psychol, 42,
304-306.

Siegel, P. S., and M. F. MacDonnell (1954) : A repetition of the Cal-
vin-Bicknell-Sperling study of conditioned drive. J. comp. physiol
Psychol, 47, 250-252.

Siegel, P. S., and M. Steinberg (1949) : Activity level as a function of
hunger. /. comp. physiol. Psychol, 42, 413-416.

Skinner, B. F. (1932): Drive and reflex strength. /. gen. Psychol, 6,
22-37.

Skinner, B. F. (1933): The measurement of spontaneous acti\ity. /.
gen. Psvchol, 9, 3-24.

REFERENCES 385

Skinner, B. F. (1938) : The behavior of organisms. New York: Apple-
ton-Century-Crofts, Inc.

Slonaker, J. R. (1912): The normal activity of the albino rat from
birth to natural death, its rate of growth and the duration of life.
/. anim, Behav., 2, 20-42.

Smith, M., R. Pool, and H. Weinberg (1959): The effect of pe-
ripherally induced shifts in water balance on eating. J. comp.
physiol. Psychol, S2, 289-293.

Smith, M. P., and P. J. Capretta (1956): Effects of drive level and
experience on the reward value of saccharine solutions. J. comp.
physiol. Psychol, 49, 553-557.

Solley, C. M., and R. Lee (1955) : Perceived size: closure versus sym-
bolic value. Amer. J. Psychol, 68, 142-144.

Solomon, R. L., and D. H. Howes (1951): Word frequency, per-
sonal values, and visual duration thresholds. Psychol Rev., 58,
256-270.

Solomon, R. L., L. J. Kamin, and L. C. Wynne (1953): Traumatic
avoidance learning: the outcomes of several extinction procedures
with dogs. /. abnorm. soc. Psychol, 48, 291-302.

Solomon, R. L., and L. C. Wynne (1953) : Traumatic avoidance learn-
ing: acquisition in normal dogs. Psychol Monogr., 67, No. 4
(Whole No. 354).

Solomon, R. L., and L. C. Wynne (1954): Traumatic avoidance
learning: the principles of anxiety conservation and partial ir-
reversibility. Psychol Rev., 61, 353-385.

Sorokin, P. A. (1942) : Men and society in calamity. New York: E. P.
Button & Co., Inc.

Spence, K. W. (1944) : The nature of theory construction in contem-
porary psychology. Psychol. Rev., SI, 47-68.

Spence, K. W. (1948) : The postulates and methods of 'behaviorism.'
Psychol. Rev., SS, 67-78.

Spence, K. W. (1951d): Theoretical interpretations of learning. In
C. P. Stone (ed.). Comparative psychology. Englewood Cliffs,
N.J.: Prentice-Hall, Inc.

Spence, K. W. (1951b): Theoretical interpretations of learning. In
S. S. Stevens (ed.). Handbook of experimental psychology. New
York: John Wiley & Sons, Inc.

Spence, K. W. (1956): Behavior theory and conditioning. New
Haven, Conn.: Yale University Press.

Spence, K. W. (1958) : A theory of emotionally based drive (D) and
its relation to performance in simple learning situations. Amer.
Psychologist, 13, 131-141.

Spence, K. W., and I. E. Farber (1953) : Conditioning and extinction
as a function of anxiety. /. exp. Psychol, 45, 116-119.

386 THE MOTIVATION OF BEHAVIOR

Spence, K. W., I. E. Farber, and E. Taylor (1954): The relation of
electric shock and anxiety to level of performance in eyelid con-
ditioning. /. exp. Psychol, 48, 404-408.

Spence, K. W., and L. E. Ross (1957) : Experimental evidence on the
relation between performance level in eyelid conditioning and
anxiety (drive) level. Technical Report 5, Studies of influence
of motivation on performance in learning, Contract N9 onr
93802, Project NR 154-107, Office of Naval Research.

Spence, K. W., and J. A. Taylor (1951) : Anxiety and strength of the
UCS as determiners of the amount of eyelid conditioning. J. exp.
Psychol, 42, 183-188.

Stein, K. B. (1953): Perceptual defense and perceptual sensitization
under neutral and involved conditions. /. Pers., 21, 467-478.

Stellar, E. (1954): The physiology of motivation. Psychol Rev., 61,
5-22.

Stevens, S. S. (1946): The science of noise. Atlantic Monthly, 178,
96-102, July, 1946.

Stevens, S. S. (1951): Mathematics, measurement, and psycho-
physics. In S. S. Stevens (ed.), Handbook of experimental psy-
chology. New York: John Wiley & Sons., Inc.

Strouthes, A., and H. C. Hamilton (1959): Fear conditioning as a
function of the number and timing of reinforcements. J. Psychol,
48, 131-139.

Sullivan, J. J. (1950) : Some factors affecting the conditioning of the
galvanic skin response. Unpublished doctoral dissertation, State
University of Iowa.

Switzer, S. A. (1930) : Backward conditioning of the lid reflex. /. exp.
Psychol, 13, 76-97.

Tandler, B. F. (1958) : Judgments of circular area and of line length
under high and low levels of motivation. (Unpublished. Personal
communication.)

Taylor, J. A. (1951): The relationship of anxiety to the conditioned
eyelid response. J. exp. Psychol, 41, 81-92.

Taylor, J. A. (1956d): Drive theory and manifest anxiety. Psychol.
Bull, 53, 303-320.

Taylor, J. A. (1956b): Physiological need, set, and visual duration
threshold. /. abnorm. soc. Psychol, S2, 96-99.

Taylor, J. A., and K. W. Spence (1952) : The relationship of anxiety
level to performance in serial learning. /. exp. Psychol, 44, 61-64.

Teel, K. S. (1952) : Habit strength as a function of motivation during
learning. /. comp. physiol Psychol, 4S, 188-191.

Teel, K. S., and W. B. Webb (1951) : Response evocation on satiated
trials in the T-maze. /. exp. Psychol, 41, 148-152.

Thorndike, E. L. (1898): Animal intelligence; an experimental study
of the associative processes in animals. Psychol. Rev. Monogr.
Suppl, 2, No. 4 (Whole No. 8).

REFERENCES 387

Thorndike, E. L., and I. Lorge (1944): The teacher's word book of

30,000 words. New York: Teachers College, Columbia University.

Tolmaii, E. C. (1932) : Purposive behavior in animals and men. New

York: Appleton-Century-Crofts, Inc.
Tolman, E. C. (1942) : Drives toward war. New York: Appleton-Cen-
tury-Crofts, Inc.
Tolman, E. C. (1948): Cognitive maps in rats and men. Psychol.

Rev., SS, 189-208.
Tolman, E. C. (1949) : There is more than one kind of learnmg. Psy-
cho?. Rev., 56, 144-155.
Tolman, E. C. (1951): A psychological model. In T. Parsons and
E. A. Shils (eds.). Toward a general theory of action. Cambridge,
Mass.: Harvard University Press.
Tolman, E. C. (1952) : A cognition motivation model. Psychol. Rev.,

59,' 389-400.
Tolman, E. C, and H. Gleitman (1949): Studies in spatial learn-
ing: VII. Place and response learning under different degrees of
motivation. J. exp. Psychol, 39, 653-659.
Tolman, E. C, C. S. Hall, and E. P. Bretnall (1932) : A disproof of
the law of effect and a substitution of the laws of emphasis, mo-
tivation, and disruption. J. exp. Psychol, IS, 601-614.
Tolman, E. C, and L. Postman (1954): Learning. In C. P. Stone
(ed.). Annual review of psychology. Palo Alto, Calif.: Annual
Reviews, Inc.
Troland, L. T. (1928): The fundamentals of human motivation.

Princeton, N.J.: D. Van Nostrand Company, Inc.
Veroff, J., S. Wilcox, and J. W. Atkinson (1953): The achievement
motive in high school and college age women. J. abnorm. soc.
Psychol, 48, 108-119.
Verplanck, W. S., and J. R. Hayes (1953) : Eating and drinking as a
function of maintenance schedule. J. comp. physiol Psychol, 46,
327-333.
Wagner, A. R. (1959): The role of reinforcement and nonremforce-
ment in an "apparent frustration effect." /. exp. Psychol, SI, 130-

136.
Warden, C. J. (1931): Animal motivation. Experimental studies on

the albino rat. New York: Columbia University Press.
Warner, L. H. (1927) : A study of sex behavior in the white rat by

means of the obstruction method. Comp. Psychol Monogr., 4,

No. 5 (Whole No. 22).
Webb, W. B. (1949) : The motivational aspect of an irrelevant drive

in the behavior of the white rat. J. exp. Psychol, 39, 1-14.
Welker, W. I. (1956): Effects of age and experience on play and

exploration of young chimpanzees. J. comp. physiol. Psychol, 49,

223-226.
Wenar, C. (1954): Reaction time as a function of manifest anxiety

388 THE MOTIVATION OF BEHAVIOR

and stimulus intensity. /. abnorm. soc. Psychol., 49, 335-340.

White, C. T., and H. Schlosberg (1952): Degree of conditioning of
the GSR as a function of the period of delay. /. exp. Psychol, 43,
357-362.

Whiting, J. W. M., and I. L. Child (1953): Child training and per-
sonality: a cross-cultural study. New Haven, Conn.: Yale Univer-
sity Press.

Whittaker, E. M., J. C. Gilchrist, and J. W. Fischer (1952): Per-
ceptual defense or response suppression. /. abnorm. soc. Psychol.,
47, 732-733.

Wiener, M. (1955): Word frequency or motivation in perceptual
defense. /. abnorm. soc. Psychol., SI, 214-218.

Williams, J. E. (1955): Mode of failure, interference tendencies and
achievement imagery. J. abnorm. soc. Psychol., SI, 573-580.

Williams, M. (1950): The effects of experimentally induced needs
upon retention, /. exp. Psychol., 40, 139-151.

Wispe, L. G., and N. C. Drambarean (1953): Physiological need,
word frequency, and visual duration thresholds. /. exp. Psychol.,
46, 25-31.

Wolfe, J. B. (1936): Effectiveness of token-rewards for chim.panzees.
Comp. Psychol. Monogr., 12, No. 5 (Whole No. 60).

Woodworth, R. S. (1918): Dynamic psychology. New York: Colum-
bia University Press.

Wycoff, L. B., Jr. (1952): The role of observing responses in dis-
crimination learning. Part I. Psychol. Rev., 59, 431-442.

Yamaguchi, H. G. (1951 ) : Drive (D) as a function of hours of hunger
(h). J. exp. Psychol, 42, 108-117.

Yamaguchi, H. G. (1952) : Gradients of drive stimulus (S^) intensity
generalization. /. exp. Psychol, 43, 298-304.

Yerkes, R. M., and J. D. Dodson (1908) : The relation of strength of
stimulus to rapidity of habit formation. /. comp. Neurol, 18,
459-482.

Young, P. T. (1936) : Motivation of behavior. New York: John Wiley
& Sons, Inc.

Young, P. T. (1949): Food-seeking drive, affective process, and learn-
ing. Psychol Rev., S6, 98-121.

Young, P. T. (1955) : The role of hedonic processes in motivation. In
M. R. Jones (ed.), Nebraska symposium on motivation. Lincoln,
Nebr.: University of Nebraska Press.

Zatskis, J. (1949): The effect of the need for achievement on lin-
guistic behavior. Unpublished master's thesis. Wesleyan Univer-
sity. Cited in McClelland, Atkinson, Clark, and Lowell, 1953.

Zeaman, D., and N. Wegner (1954): The role of drive reduction in
the classical conditioning of an autonomically mediated response.
J. exp. Psychol, 48, 349-354.

NAME INDEX

llllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllli:

Adams, }., 321

Adamson, J. F., 315

Adolph, E. F., 68, 72

Aiken, E. G., 346

Amsel, A., 83, 84, 108,132, 133, 1

156, 163, 195, 205, 211-213
Anand, B. K., 66
Anderson, E. E., 187
Arakelian, P., 80
Arenberg, D., 283
Armus, H. L., 167
Aronson, E., 250
Ashley, W. R., 295, 297
Asimow, A., 91
Atkinson, J. W., 131, 180, 184, 2

239, 241-244, 246, 248, 282

Bahrick, H. P., 132
Bakan, D., 3
Baker, R. A., 76
Baldwin, M. V., 226
Barnes, G. W., 358
Barry, H., Ill, 120
Bass, B., 78, 120
Beams, H. L., 298
Bechtoldt, H. P., 274, 275
Beebe-Center, J. G., 24
Bendig, A. W., 260
Bergmann, G., 46
Berkun, M. M., 65, 338
Berlvne, D. E., 331, 334, 335
Bevan, W., Jr., 297
Bicknell, E. A., 189

Bills, A. G., 185
Bindra, D., 131, 258
Birch, D., 178, 179, 187
Bitterman, M. E., 256, 315, 344
55, Bolles, R. G., 131

Boren, }. J., 132, 353, 358

Boslov, G. L., 348

Bowden, J., 349

Bower, G. H., 86, 353, 354

Brady, J. V., 132, 163, 168, 353, 354,

358
Brandauer, C. M., 122
Brantley, J. J., 132, 156
Braunstein, E. P., 19
27, Bretnall, E. P., 232
Bricker, P. D., 308
Broadhurst, P. L., 85
Brobeck, J. R., 66, 69
Bronson, W. C., 314
Brown, D. R., 321
Brown, G. W., 86, 351, 354, 356
Brown, J. S., 70, 108, 148, 150, 151,

156, 157, 166, 172, 182, 195,

201, 203, 208, 249, 282, 316,

337, 357
Brown, M. L., 86, 351, 356
Brown, P. A., 214
Brown, W. L., 208, 219
Brozek, J., 226

Bruner, J. S., 289, 293, 298, 310, 320
Bugelski, R., 148
Burnstein, E., 178, 187
Burros, R. H., 168
Bursten, B., 351, 353, 356
Butler, R. A., 331-333, 337, 338
389

390

THE MOTIVATION OF BEHAVIOR

Calvin, J. S., 189

Campbell, B. A., 83, 131

Campbell, N. R., 51, 54

Cannon, W. B., 63, 64, 144, 147

Capretta, P. J., 69

Carlton, P. L., 90

Carpenter, J. A., 118, 131

Carter, L. F., 293, 295

Castaneda, A., 262, 263

Champion, R. A., 90

Chapanis, A., 235, 308

Chapman, R. M., 338

Chein, I., 284

Child, I. L., 131, 220, 261

Chiles, W. D., 350

Chodorkoff, B., 323

Clark, R. A., 131, 178, 180, 184,

187, 239, 241, 242, 245, 253
Clark, R. E., 279, 280
Cohen, B. D., 86, 351, 354, 356
Cohen, M. R., 48, 51
Collier, C, 131, 202, 205, 218
Conger, J. J., 168
Connor, T. J., 133
Conrad, D., 132, 353, 358
Cooper, J. B., 219
Correll, R. E., 350
Cotton, J. W., 80, 130
Courts, F. A., 42, 132, 185
Cowles, }. T., 174
Crespi, L. P., 178, 208
Crutchfield, R. S., 228

Daniel, W. J., 74, 102, 111
Davenport, J. W., 78, 90
Davis, J. D., 123, 132
Davis, M., 357
Davis, R. C, 5
Davis, R. H., 120, 131
Davitz, J. R., 347
Deese, J., 118, 131, 233, 236
Delgado, J. M. R., 351, 353, 356
Delhagen, J. E., 358
Dember, W. N, 331
Dinsmoor, J. A., 91, 125
Dodson, }. D., 91

Dollard, }., 87, 176, 180, 182, 214,
343

Doob, L. W., 214
Drambarean, N. C, 284
Duffy, E., 108
Dukes, W. F., 297

Eisman, E., 91, 93

Eriksen, C. W., 322

Estes, W. K., 130, 131, 163, 167

Farber, I. E., 70, 108, 131, 132, 148,
151, 152, 156, 166, 168, 195,
201, 203, 208, 233, 236, 237,
248, 257, 258, 260, 321, 357

Fernandez, M. B., 133

Finch, C, 201, 208

Fischer, J. W., 314

Fiske, D. W., 320

Fonda, C. P., 322

Forgays, D. C, 358

Fowler, H., 331

Franks, C. M., 229

Fredenburg, N. C, 77, 79, 126, 128

Freeman, J. T., 314

French, E. C, 250

French, }. R. P., Jr., 195

Freud, S., 110, 144, 169

Fuster, J. M., 132, 350

Galambos, R., 168

Gardiner, H. M., 24

Garner, W. R., 235

Gentry, G., 208, 219

Gerall, A. A., 348

Ghent, L., 76

Gilbert, R. W., 233

Gilchrist, J. C, 285, 286, 288, 296,

314
Gilhousen, H. C, 219
Glanzer, M, 331
Gleitman, H., 90
Goodman, C. C, 289, 293
Greene, }. E., 121
Grice, G. R., 115, 116, 123
Cropper, G. L., 314
Grossman, M. I., 64-66, 68
Guetzkow, H., 226

NAME INDEX

391

Guthrie, E. R., 1, 105, 130, 218, 339
Gwinn, G. T., 160, 162

Haigh, G. v., 320

Hall, G. S., 3, 232

Hall, J. F., 132

Hamilton, G. V., 208

Hamilton, H. G., 347

Hancock, W., 213

Haner, G. F., 214

Hanford, P. V., 132

Harlow, H. F., 329, 332, 333, 337

Harper, R. S., 295, 297

Hayes, }. R., 134

Heath, R. G., 354

Heathers, G. L., 80

Hebb, D. O., 110, 131, 349

Hedlund, J. L., 246, 252

Henderson, R. L., 358

Hilden, A. H., 5

Hilgard, E. R., 208, 256, 259

Hillman, G., 120, 131

Holder, E., 131, 202, 205, 218, 219

Holder, W. B., 131, 202, 205, 218

Hollenberg, E., 219

Holton, R. B., 214, 216, 217

Holtzmann, W. H., 256

Horenstein, B. R., 73, 74, 80

Horowitz, L., 358

Horton, G. P, 7, 218

Hovland, G. I., 13, 14, 208

Howes, D. H., 129, 311, 320, 321

Hughes, J. B., II, 260

Hull, G. L., 33, 57, 60, 99, 116,

117, 176, 195, 203, 329, 339,

348
Hunt, H. F., 155, 163, 168
Hunter, W. S., 120, 131
Hurley, J. R., 252

Ingram, W. R., 350

Jacobs, A., 157, 159, 182, 195
Janowitz, H. D., 68
Jasper, H. H., 348
Jenkin, N., 299

Jerome, E. A., 133
Johanson, A. M., 132
Johnston, R. A., 252
Jones, L. V., 256
Jones, M. C, 168

Kabrick, R. P., 156

Kagan, J., 338

Kalish, H. I., 148, 151, 160-162, 166,

167
Kamin, L. J., 166, 169
Kaplan, M., 87
Kaplan, S. J., 256
Keenan, J., 233
Kendler, H. H., 134
Kessen, M. L., 65
Ketchel, R. G., 87
Kimble, G. A., 78, 120, 131
Kish, G. B., 332, 357, 358
Klein, G. S., 298
Kling, J. W., 358
Kniffen, G. W., 315
Knott, J. R., 350
Koch, S., 74, 80, 102, 111
Kraeling, D., 83, 131
Krauskopf, J., 344

Lacy, O. W., 315

Lambert, W. W., 298, 301

Lamoreaux, R. R., 162

Lawrence, D. H., 76, 160, 162, 166

Lawson, R., 195

Lazarus, R. S., 233, 236, 283, 306,

322
Lee, R., 297
Levin, H., 358
Levine, R., 284
Levy, N., 338
Lewin, K., 57, 194
Lewinger, N., 315
Libby, A., 167
Liberman, A. M., 245
Lindsley, D. B., 110, 348-351
Lindzey, G., 3, 24
Loess, H. B., 78, 90, 112, 120, 131
Logan, F. A., 5, 88
Lorge, I., 312

392

THE MOTIVATION OF BEHAVIOR

Lowell, E. L., 131, ISO, 184, 219,

239, 242-245, 252
Ludvigson, W. H., 275, 276
Lysak, W., 296, 308

McAllister, W. R., 234

McCandless, B. R., 262, 263

McCleary, R. A., 306

McClelland, D. C, 59, 131, 176,
180, 184, 227, 239, 241, 242,
245, 248, 250, 253, 282

MacCorquodale, K., 246

MacDonnell, M. F., 190

McDonough, J. S., 274, 275

McDougall, W., 110

McGinnies, E., 305, 307, 312, 313,
320

Madsen, K. B., 24

Magoun, H. W., 349

Malmo, R. B, 109, 350

Maltzman, I., 91, 132, 156

Marquis, D. G., 208, 259

Marx, M. H., 59, 131, 195, 202, 205,
218, 358

Marzocco, F. N., 156, 208, 210

Mason, W. A., 76

Masserman, }. H., 356

Mausner, B., 298

Meehl, P. E., 246

Meister, D. E., 298

Melton, A. W., 198

Meryman, J. J., 133, 152-154, 156,
167

Metcalf, R. C, 24

Meyer, D. R., 93, 131

Miller, N. E., 33, 65, 86, 87, 90, 109,
121, 132, 142, 148, 156, 160,
162, 166, 176, 180, 182, 189,
195, 208, 214, 248, 316, 331,
343, 345, 351, 353, 354, 357

Miller, W. C, 121

Milner, P., 352

Miner, R. W., 63, 64

Moeller, G., 345

Montgomery, K. C., 331-333, 338

Moody, J. A., 133

Morgan, C. T., 63, 68, 70, 109, 235

Moruzzi, G., 349

Mowrer, O. H., 70, 144, 162, 167,

182, 214, 340, 344, 346, 357
Muenzinger, K. F., 232
Murfin, F. L., 167
Murphy, D. B., 308
Murphy, G., 284, 301
Murray, H. A., 184
Myers, A. K., 189, 331
Myers, T. I., 94

Nagaty, M. O., 132

Nagel,'E., 48, 51

Nauta, W. J. H., 354

Neel, A. F., 323

Nesberg, L. S., 285, 286, 288

Nissen, H. W., 3

Noble, M. E., 131

Olds, }., 352-355, 358
Olmstead, D. L., 5
Osgood, C. E., 70
Osier, S. F., 236
Otis, L. S., 155

Palermo, D. S., 262, 263

Passey, G. E., 234

Paterson, A. L., 258

Pavlov, I. P., 5, 171

Perin, C. T., 80

Pool, P., 65

Postman, L., 129, 228, 232, 246, 298,

310, 314, 320, 321
Powloski, R. F., 134
Proshansky, H., 301
Pustell, T. E., 309

Ramond, C. K., 78, 90, 104
Reece, M. M., 308
Reed, P., 344
Reese, T. W., 51, 52
Reynolds, R. W., 353
Richter, C. P., 103
Riesen, A. H., 13, 14
Roberts, C. L., 358
Roberts, W. W., 86, 351, 353, 354,
358

NAME INDEX

393

Rodriques, }. S., 293
Rohrer, J. H., 195
Rosenbaum, G., 253, 257
Rosenthal, B. G., 294
Rosner, B. S., 5
Ross, L. E., 255
Rossi, G. F., 348
Roussel, }., 195, 211-213
Runyon, D. L., 295, 297

Saltz, E., 115, 116

Saltzman, I., 74, 80

Sampliner, R. I., 65

Sampson, P. B., 348

Samuels, I., 351

Sanford, R. N., 227, 284

Schlesinger, H. }., 298

Schlosberg, H., 340, 345, 350

Schooler, K., 293, 295

Schulman, A., 353

Schwartz, R. D., 5

Sears, R. R., 214

Segall, M., 331

Seward, J. P., 59, 205

Sheffield, V. F., 195

Sheriffs, A. C., 251

Sherman, H., 305, 313

Shirley, M., 187

Sidman, M., 132, 353, 358

Siegel, A., 298

Siegel, P. S., 71, 72, 90, 132, 156,

188, 190
Skinner, B. F., 6, 32, 33, 120, 163,

167, 186, 208, 340
Slonaker, }. R., 187
Smith, M., 65
Smith, M. P., 69
Solley, C. M., 297
Solomon, L. N., 167, 344
Solomon, R. L., 129, 166, 167, 169,

298, 301, 304, 309, 311, 320,

321
Sorokin, P. A., 226
Spence, K. W., 14, 17, 23, 46, ",

87, 90, 105-107, 109, 132, 134,

177, 195, 233, 234, 253, 255-

262, 329, 340
Sperling, D. S., 189

Sperry, M., 219
Sprague, }. L., 260
Stebbins, W. C., 168
Stein, K. B., 523
Steinberg, M., 132
Stellar, E., 63, 110
Stevens, C. M., 5
Stevens, S. S., 48, 235
Stevenson, S. S., 195, 2(
Strouthes, A., 347
Strzelecki, J., 258
Sullivan, J. J., 544
Switzer, S. A., 208

Tandler, B. F., 279, 280
Taylor, E, 152, 255, 257, 258
Taylor, }. A., 108, 255-257, 260, 285
Teel, K. S., 89, 105
Thorndike, E. L., 208, 312
Tolman, E. C., 7, 28, 57, 70, 90,

108, 175, 232, 246, 247
Troland, L. T., 24

Veroff, J., 246
Verplanck, W. S., 134

\\^agner, A. R., 213
^\'arden, C. }., 72
Warner, L. H., 17
Waterhouse, I. K., 151, 261
Watson, P. D., 298, 501
Webb, W. B., 89, 122
Wegner, N., 344
Weinberg, H., 65
Welker, W. I., 555
W^enar, C., 258
White, C. T., 545
^^'hiting, J. W. M., 219
Whittaker, E. M., 514
Wiener, M., 313
Wilcox, S., 246
\\'illiams, J. E., 252
\\'illiams, M., 230
Wispe, L. G., 284
\\^olfe, J. B., 174
\\'oodrow. P., 65

394

THE MOTIVATION OF BEHAVIOR

Woodworth, R. S., 110
Wycoff, L. B., Jr., 274
Wynne, L. C, 166, 167, 169, 304

Yamaguchi, H. G, 81, 133
Yerkes, R. M., 91

Young, P. T., 24, 59, 76
Yousem, H., 283

Zanchetti, A., 348
Zatskis, J., 242
Zeaman, D., 344

SUBJECT INDEX

Achievement motivation:

in addition-task performance, 243,

244
alternative methods of definition,

250-253
in anagrams task performance, 244,

245
effects on behavior, 242-246
and intelligence, 244, 245
and Rorschach scores, 242
and tachistoscopic recognition

thresholds, 245, 246
Acquired drive:

affective-arousal model of, 180-

182
associative-associative theory of,

140, 141
associative-motivational theory of,

140, 141
conditioned fear (anxiety) as, 144-

176
{See also Fear)
effects of primary sources of drive

on, 141, 142
problem of, 139-144
and response-produced stimuli, 180
Tg-s mechanism of, 176-180
verbal commands as sources of,

182-186
(See also Drive, secondary

sources of)

Activation, 109, 110
Additivity, property of, 54
Affective arousal:
and fear, 182

as motivating agency, 239, 240
motivating role of, 180-182
and Tg-Sg mechanism, 182
Aggressiveness and perception, 322,

323
Air deprivation as motivating variable,

85, 86
Air-puff intensity and eyelid condition-
ing, 234
Anticipatory goal responses {Tg-Sg

mechanism), 176-180
Anticipatory seeing responses, 337
Anxiety:

and acquired drive for money, 171-

176
defined by manifest anxiety scale,

253, 254
drive function of, in exploratory

behavior, 337, 338
test defined, in children, 262,
263
as chronic state, 257-259
and complex tasks, 259-263
and distractibility, 260, 261
in eyelid conditioning, 254-257
and GSR conditioning, 256,
257

SUBJECT INDEX

395

Anxiety (continued) :

and relative habit strength, 259-
263
Aphagia, 66

Apparent size as function of value,
289-295
(See also Perception; Size estima-
tion; Value )
Ascending reticular activating system

(see Reticular system)
Associative-associative theory of ac-
quired drive, 140, 141
Associative interpretations, 110-113
advocates of, 129-131
defined, 98, 110, 111
generalization version, 115-126
and choice behavior, 123-126
and deprivation-time shifts, 118-

121
and need-state changes, 121-123
habit-change version, 126-129
and choice behavior, 128-129
and deprivation shifts, 126-129
and need-state shifts, 128
probability-learning version, 1 30
stimulus-element version, 111-115
and choice behavior, 114-115
and deprivation shifts, 111-114
and need-state shifts, 114
varieties of. 111
Associative-motivational theory of ac-
quired drive, 140, 141
Associative theories (see Associative

interpretations)
Asymmetry in logic of measurement,

51
Auditory stimuli:

drive-arousing value of, and explora-
tion, 337
rewarding function of, 332

Backward conditioning in emotional

conditioning, 347
Bar pressing:

and deprivation shifts, 102, 103
inhibited by fear, 163
reinforced, by fear reduction, 157
by intracranial stimulation, 352-
356

Behavior:

general methods of studying, 21-23
and primary sources of drive in

animals, 70-94
as subject matter of psychology, 1, 2
(See also specific types of behavior)
Behavior director, habit strength as,

58, 59, 61
Behavioral approach, 22, 23
Behavioral variability and concept of

purpose, 7
Body -weight deficit:

and discrimination, 91-93
and water consumption, 72
Brain stimulation (see Intracranial
stimulation)

Cardiac conditioning and UCS dura-
tion, 345
Central motive state (c.m.s.), 109,

110
Choice behavior:

and deprivation, generalization in-
terpretation of, 123-126
motivational theory of, 103, 104
and food deprivation, habits
equated, 89, 90
habits unequal, 90
and primary motivational variables,
87-90
Choice box, two-bar, 88
Coins, apparent size of, 289-295
CS-UCS interval in emotional con-
ditioning, 345
Conditioned hunger (see Hunger)
Conditioning:

and anxiety level (see Anxiety)
dual role of unconditioned stimuli

in, 233, 234
of fear (see Fear conditioning)
(See also Emotional conditioning)
Confinement, escape from, in explora-
tory behavior, 338, 339
Conflict as source of drive, 178, 219,
220
in primitive societies, 220
(See also Frustration behavior;
Thwarting)

396

THE MOTIVATION OF BEHAVIOR

Consummatory behavior:

and deprivation time, 71-76
drive-reducing function of, 68
effects of experience on, 76
as estimate of drive, 75
and internal stimuli, 75, 76
regulation of, 62-66

by gastrointestinal factors, 64-66
by oropharyngeal stimulation,

64-66
role of nutrients in, 66
and sham feeding, 65
Contiguity in fear conditioning, 339-
348

Data {see Psychological data)
Defecation as index of conditioned

fear, 155
Definitions:

operational, 44-46
significant, 46, 47
Delay of reinforcement in emotional

conditioning, 343
Demands, 108
Deprivation history in defining drive,

35-37
Deprivation schedules as determinants

of drive, 93
Deprivation time {see Air deprivation;
Food deprivation; Hunger;
Thirst)
Derived measurement {see Drive,

quantification of)
Difference limens and motivational

variables, 278, 279
Discriminatory behavior and primary

motivational variables, 90-94
Disproportionality of stimulus and re-
sponse, 29, 30
Drinking and postshock emotionality,

156
Drive:

acquired {see Acquired drive)
conflicting definitions of, 67, 68, 72
curiosity, 328

as distinguished from drives, 59, 60
as distinguished from needs, 67-70
relation to drive-reduction hy-
pothesis, 69

Drive {continued) :
emotion-produced, 178
exploratory, 328

circular definitions of, 334
externalized, 187, 188
frustration {see Frustration drive)
gregariousness, 175, 176
hunger-produced, as affected by

rhythmical feeding, 178, 179
manifest anxiety scale, definition of,
253, 254
{See also Manifest anxiety scale)
manipulative and selective defini-
tions of, 238, 239
manipulatory, 328
multiplicative theory of {see Multi-
plicative-drive theory )
perceptual curiosity, 335, 336

and Tg-Sg mechanism, 335, 336
primary sources of, 61, 62
and choice behavior, 87-90
in discriminatory behavior, 90-94
and instrumental behavior, 76-87
in perception, 282-289
quantification of, 47-54
bv counting, 49, 50
and derived measurement, 54
and fundamental measurement,

53, 54
general rules for, 47-49
by ranking, 50-53
response-defined, in humans, 237-

262
secondary sources of, 62

compared with primary sources,

62
defined, 139

and perception, 289-323
{See also Acquired drive)
and stimulus intensity, 180
thwarting-produced, 194-196
visual exploratory, difficulties with,

336-338
weight-loss definition of, 91-92
Drive reduction:

and consummatory activity, 68
and saccharin, 69
and secondary reinforcement, 70
Drive stimuli, 75

SUBJECT INDEX

397

Eating and postshock emotionality,

156
Electric shock:

associative consequences, 232, 233
competing responses induced by,

232, 233
and escape behavior, 83, 84
and human performance, 231-233
informative effects of, 232
in instrumental behavior, 82-85
intracranial, 86, 87
motivating and directing functions

of, 58, 59
motivating effects of, in animals,

82, 86
persisting effects of, 232, 233
punishing effects of, 231, 232
randomly administered, 233
reduction of, and perceptual thresh-
olds, 308, 309
Electroconvulsive shock and con-
ditioned fear, 163, 168
Emotion of hope as drive source,

182
Emotional conditioning:
backward order in, 347
contiguity plus drive-reduction view

of, 340-348
critical variables in, 341, 342
distinguished from mechanisms,
341, 342
and CS-UCS interval, 345
and delay of reinforcement, 343
"different-process" theories of, 340,

341
by drive reduction, 340
experimental designs, 342-348
intermittent shock in, 345
and pupillary conditioning, 348
"same-process" theories of, 339,

340
temporal intervals in, 347
and UCS duration, 342-346
cardiac conditioning and, 345
Emotional responses, role as motiva-
tors, 107
Emotionahty:

associative effects of, and percep-
tion, 305, 306

Emotionality {continued):

contribution to drive, and percep-
tion, 305
postshock, compared with fear, 155,

156
stimulus-produced, and perception
{see Perception)
Energy mobilization, 109
Eyelid conditioning:

and air-puff intensity, 234, 256
anxiety and UCS intensity in, 255,
256
Excitatory potential, determinants, 99
Expectancy (r^-s^ mechanism), 176-

180
Experience, role of, in drive definition,

35-37
Exploratory behavior :

and drive-reduction theory, 337,

338
escape from confinement in, 338,

339
extinction of, 332
and fear, 333, 334
generalization of, 332
and nonexploratory behavior, 334
studies of, 331-333
Exploratory drive {see Drive)
Extensive dimension, 53, 54
External stimuli, motivational attri-
butes of, 327, 328
Extinction of fear, significant variables
in, 167, 168

Facihtation, selective, of responses,

248-250
Factor X as hypothetical behavior de-
terminant, 31-41
Familiarity with words in perceptual

recognition, 311-315
Fear:

acquisition of, and startle responses,

149-151
and alcohol, 168
and audiogenic seizures, 168
and bar pressing, 163
combined with hunger, 152-154
and defecation, 155

398

THE MOTIVATION OF BEHAVIOR

Fear (continued) :

distinguished from residual emo-
tionality, 155, 156
and electroconvulsive shock, 163,

168
as energizer, 147-155
extinction of, and startle responses,
151
variables affecting, 167, 168
in human subjects, and GSR, 154,

155
in human subjects, higher-order
conditioning of, 170, 171
as learned response, 169-171
role of language, 170, 171
(See also Anxiety)
inhibition of drinking by, 163, 164
as inhibitor and energizer, 165-166
inhibitory effects of, 162-166
procedures for estimating strength

of, 145-147
reduction, and extinction trials, 161
and fear-conditioning trials, 160,

161
as reinforcement, 156-162
spontaneous recovery of, and startle

responses, 151, 152
and startle responses, 148-154
variables affecting strength of, 166-
168
Fear conditioning:

contiguity in, 339-348
procedures, 144, 145
theories of, 139, 142
(See also Emotional conditioning)
Fistulas, esophageal, and sham-feed-
ing, 65
Food deprivation:

facihtative effects of, 132
and human performance, 225-230
and imaginative responses, 241
and perception, 282-289
and running speed, 77-79
{See also Hunger)
Fractional anticipatory goal reactions

{Tg-Sg mechanism), 176-180
Frustration :

antecedents of, 203, 204
anticipator)', 205-207

Frustration (continued) :
as aversive condition, 206
conditioned, 205-207
conflict-produced, 203, 204
as drive source, 203-207

(See also Thwarting)
effects on behavior, 203, 204
as implicit reaction, 205
from nonreward, 203, 205
as stimulus source, 203, 204, 206
unlearned reactions to, 204

Frustration behavior:

associative conceptions of, 200-202
and extinction, 202
motivational theories of, 202-207
and stimulus similarity, 201
and transfer of training, 202
(See also Thwarting)

Frustration drive:

associative conceptions of, 210,

211
and bar pressing, 208-211
and delay, 218, 219
and deprivation time, 209, 210
as energizer, 207-217
and goal distance, 214-217
inhibitory effects of, 218, 219
and massing of extinction trials,

209, 210

reduction of, as reinforcement, 218
and reinforcement frequency, 209,

210, 216

and response vigor (children), 214-

218
and running speed in rats, 211-
214
Frustration stimuli, inhibitory proper-
ties of, 206
Function, stepwise, 29, 30
Functional relations, 13-21
OR, 17, 72
OS-R, 20
R-R, 14, 240, 241
and reliability, 15
and validity, 16, 17
SO-R, 19
S-R, 13
Fundamental measurement (see
Drive, quantification of)

SUBJECT INDEX

399

Galvanic skin response:

conditioning of, and anxiety, 256,
257

in defining subception, 306, 307
Generalization :

on drive-stimulus dimension, 178,
179

of habit strength, 116-118

{See also Stimulus generalization)
Gregariousness, drive of, 175, 176

Habit strength:

as aflPected by stimulus impoverish-
ment, 281
contribution of, to excitatory poten-
tial, 99, 100
directive functions of, 58, 59, 61
generalized, defined, 116-118
inferred from psychophysical data,
273-274
Hunger:

conditionabihty of, 186-191
"conditioned," and food consump-
tion, 190
and maze performance, 187, 188
and response acquisition, 189
and running-v/heel activity, 187
in discriminatory behavior, 91-93
effect of, on need-related responses,
227-229
on startle responses, 152-154
and fear, combination of, 152-154
and food consumption, 73, 74
and human performance, 225-230
and imaginal processes, 227-229
in instrumental behavior, acquisi-
tion of, 76-79
local theory of, 63-64
and resistance to extinction, 79-81
and retention of verbal materials,

230
and set, 228, 229
stimulus control of, 66, 70
and verbal food-response proba-
bility, 228, 229
and word completion, 228, 229
{See also Consummatory behavior;
Discriminatory behavior; In-
strumental behavior)

Hurdle-jumping, reinforcement of, by
fear reduction, 157-162

Hyperphagia, 66

Hypnosis in value-size judgments, 295,
296

Hypothalamus:
and eating, 66
as locus of central motive state, 110

Hypothetical constructs {see Inter-
vening variables)

Illuminance matching:

constant errors in, and drive, 287,

288
incentive motivation in, 287, 288
interpreted, 286-288
role of thirst in, 285-289
time errors in, 285-288
Incentive motivation (K):
defined, 107, 108
as exteroceptively aroused drive,

329, 330
and valuable objects, 290, 291

group differences in, 291
{See also Tg-Sg mechanism)
Instructions:

cue functions of, 236, 237
motivating effects of, 235-237
success and failure, 236, 237
{See also Motivating instructions)
Instrumental behavior:

acquisition of, and deprivation time,

77, 78
and air-deprivation time, 85, 86
and deprivation time, 76-82
and electric shock, 82-85
extinction of, and deprivation time,

79-82
and intracranial stimulation, 86, 87
{See also Intracranial stimula-
tion)
noxious stimulus intensity and, 82-
87
Insulin, 69

Intensive dimension, 53
Interaction in empirical laws, 20
Intermediary concepts {see Interven-
ing variables)

400

THE MOTIVATION OF BEHAVIOR

Intervening variables, 23, 28-47
defined circularly, 34, 35
independently defined in terms, of
antecedent conditions, 35-37
of organic variables, 37, 38
of stimulus conditions, 38-40
of task performance, 37
naming of, 40, 41
relations suggesting need for, 29-
34
Intracranial stimulation:

and drive-reduction hypothesis,

358, 359
effects of alerting, 350
inhibitory, 350, 351, 354
positive and negative, and in-
tensity, 354, 355
specific, 351
motivating consequences of, 86, 87,

351
nondirective aspects of, 356
and pleasure-pain centers, 353-356
reinforcing properties of, 352-356
Iowa Multiple Choice Achievement

Imagery Test, 252, 253
Iowa Picture Interpretation Test:
achievement motivation defined by,

251, 252
and performance, 252
reliabihty of, 251, 252
and TAT-defined achievement
motivation, 251, 252

Laws, guessed at, 23

{See also Functional relations)
Learning:

as distinguished from performance,

99, 100
{See also Acquired drive; Condi-
tioning; Instrumental be-
havior)
Libido need, 108

Local theory of hunger and thirst, 63,
64

Manifest anxiety {see Anxiety; Mani-
fest anxiety scale)

Manifest anxiety scale:
children's form, 262, 263
in defining drive, 253, 254
relation to psychiatric ratings, 254
relation of scores to eyelid condi-
tioning, 254-257
Manipulatory behavior, 332, 333

persistence of, 333
Masochistic behavior:
in cats, 356, 357
in rats, 357
Measurement:
derived, 54
fundamental, 53, 54
nature of, 48
varieties of, 49-54
{See also Drive, quantification of)
Method of constant stimuli and mo-
tivational variables, 272-281
Methods of data collection, 2
Molar behavior, 23
Money :

acquired drive for, and anxiety,

171-176
rewarding effects of, as anxiety re-
duction, 174, 175
as secondary reinforcement, 174
Motivating instructions:

drive-like effects on perception,

276-281
effects on recognition thresholds,

274-278
{See also Instructions; Verbal com-
mands as drive sources)
Motivation:

defined by imaginative responses,

239-253
general problem of, 24-25
response-defined, in human subjects,

237-262
{See also Drive; Motivational vari-
ables; Motives)
Motivational interpretations:
advocates of, 107-110
defined, 98

supporting evidence, 132-135
Motivational variables:
combinations of, 133, 134
criteria for identification of, 41-44

SUBJECT INDEX

401

Motivational variables {continued) :
and difference limens, 278, 279
directing and motivating functions

of, 58, 59
distinguished from steering vari-
ables, 57
effects of, interpreted associatively
(see Associative interpretations)
energizing criterion of, 41, 42
learning criterion of, 42
named b}' exclusion, 42, 43
punishment criterion of, 42
Motives :

effects of, as associative phenomena,
246, 247
on behavior, theory of, 246-248
experimental arousal of, 241, 242
interaction with cognitive expec-
tancies, 247, 248
as learned affective reactions, 239,

240
as selective facilitators, 247, 248
Motor tasks and anxiety level, 260
Multiplicative-drive theory (Hull), 60,
61, 99, 100
and absolute thresholds, 272-278
applied, 101-107
and choice behavior, 103-106
and habit-strength equality, 105,

106
and habit-strength inequality, 104,

105
implications of, 100, 101
and perception, 272-281
prediction of poorer performance

by, 105, 106
and satiation effects, 102, 103
and slope of psychophysical func-
tions, 276-279
supporting evidence, 134, 135
Muscular tension, facilitative effects
of, 132

Need-achie\'ement motive, 1 84
Needs, biological:

and consummatorv behavior, 61-
66

distinguished from dri\'e, 67-70

Noise and human performance,

235
Novel stimuli:

aversive properties of, 333
drive function of, 330-339
produced by motivation-variable

shifts, 131
reinforcing properties of, 331, 332
Noxious stimuli:

and instrumental behavior, 82-87
motivating effects on human be-
havior, 230-235
response-evoking properties of, 87
Nutrients and consummatory behav-
ior, 66

Operational definitions, 44-46
Ordinal scale {see Drive, quantifica-
tion of)

Paired-associates learning and anxi-
ety, 261, 262
Parameter in empirical laws, 20
Perception :

without awareness, 307, 308
definitional problem, 269, 270
and delayed overt responses, 268,

269
of emotion-arousing stimuli, 301-
323
and acquired-drive problem, 302
associative view of, 305, 306
and drive, 305

and linguistic behavior, 311, 312
possible mechanisms, 303-306
and stimulus distinctiveness, 303
and stimulus generalization, 302
and motivational variables, general

problem of, 271, 272
and multiplicative-drive theory,

272-281
in others, 267
and persisting emotionality, 304,

305
and personal values, 320

associative and other views of,
321, 322

402

THE MOTIVATION OF BEHAVIOR

Perception (continued) :

phenomenological conception of,

267
reinforced by shock reduction, 308,

309
of stimuli paired with shock, 306-
310
sex differences in, 309, 310
of taboo words, 310-323

associative factors in, 311-315

and familiarity, 311-315

and personality characteristics,

319-323
R-R relations in, 319, 320
and set, 314, 315
as unnecessary concept, 267, 268
value orientation and, 320-322
verbal versus autonomic definitions
of, 307, 308
Perceptual curiosity, drive of, 335,

336
Perceptual defense, 303-323

competing-response theory of, 315-

319
idiosyncratic nature of, 322, 323
Perceptual thresholds and motiva-
tional variables, interpreted asso-
ciatively, 282
Perceptual vigilance, 303-323
Physiological approach, 22
Point of subjective equality, shift with

value, 299-301
Primary needs [see Drive, primary

sources of)
Primary sources of drive and perform-
ance {see Drive)
Psychological data:
basic, 11, 12
characteristics of, 2, 10
methods of collecting, 2
Pupillary conditioning, role of pain in,
348

Quantification:

of drive {see Drive)

logical requirements of, 51, 52

R-R relations and human motive
strength, 240, 241

Tg-Sg mechanism as source of drive,

176-180
Rank-order scale {see Drive, quanti-
fication of)
Reaction time and postshock emo-
tionality, 156
Resistance to extinction:

as affected by need shifts, 122, 123
and deprivation, 74, 79-81
associative view of, 80, 81
problems of experimental design,
79-82
Response alternation, 331
Response classes:
definition of, 4-9

as actions of muscles or glands,

5
as limb movements, 5, 6
by reference to same determi-
nants, 8
in terms, of satisfaction of same
needs, 7, 8
of similar effects on environ-
ment, 6, 7
Response constancy with stimuli vari-
able, 32, 33
Response facilitation, selective, 248-
250
as apparent effect of drive (D),

248, 249
as associative phenomenon, 249,
250
Response suppression:
methods of control, 314
and recognition thresholds, 313,
314
Response variation with stimuli con-
stant, 31
Responses, definition of, 3-9

without reference to stimuli, 4
in terms of stimuli, 3, 4
Reticular system:

control of alerting function by stim-
ulation of, 351, 352
described, 348, 349
electrocortical activation by, 349
nonspecific functions of, 348-351
Running speed, food deprivation and,
77-79

SUBJECT INDEX

403

Saccharin and drive-reduction hypoth-
esis, 69
Satiation:
stimulus, 331

and T-maze performance, 89, 90
Secondary drive source {see Drive)
Secondary reinforcement, drive-reduc-
tion views of, 70
Selective facilitation {see Response

facilitation )
Self-stimulation {see Intracranial stim-
ulation; Masochistic behavior)
Sensory-change reinforcement, 357-
359
and drive-reduction hypothesis, 358,

359
and stimulus intensity, 358
Sexual behavior, drive-reduction prop-
erty of, 68
Sexual words, tachistoscopic recogni-
tion of, 322, 323
{See also Perception)
Sham feeding and need-drive distinc-
tion, 68
{See also Consummatory behavior)
Shock {see Electric shock)
Significant definitions, 46, 47
Size estimation:

as affected by rewards, 298, 299
and hypnotically induced socio-
economic status, 295, 296
and monetary value, 289-295
experimental design problems,

292
under hypnosis, 295, 296
and object value, theories of, 299-

301
of paper currency, 296, 297
of physically present coins, 293,

294
of preferred foods, 298
and socioeconomic status, 294, 295
and symbolic value, 297-299
Spatial choice situations, behavior in,
and primary drive sources, 87-
90
Stabilimeter, 148, 149
Starting speed, food deprivation and,
77-79

Startle responses:

augmented by fear, 148-154
and postshock emotionality, 156
Stimuli:

definition of, 9-12

independently of responses, 9, 11
in relation to assorted variables,

12
as subclass of physical incidents,

11
in terms of responses, 12
drive, 75

eliciting, definition of responses, 3, 4
external, as drive sources, 327, 328
intense, effects on behavior, 132
motivation-variable, 75
response-produced, as drive source,

180
{See also Novel stimuli; Noxious
stimuli )
Stimulus constancy and response vari-
ability, 30-32
Stimulus dynamism, 178
Stimulus generalization:

in associative interpretations {see

Associative interpretations)
empirical phenomenon of, 115, 116
gradient of, 115, 116
in perceptual defense theory, 317-
319
Stimulus impoverishment:
and competing habits, 281
and habit strength, 281
Stimulus novelty:

definitional problems, 334, 335
as drive source, 330-339
{See also Novel stimuli)
Stimulus satiation, 331
Stimulus variability and response con-
stancy, 32, 33
Subception, 307
Symbolic constructs {see Intervening

variables)
Sjmbolic value and size estimation,
297-299
{See also Size estimation)

T maze, performance in, and depriva-
tion, 229, 230

404

THE MOTIVATION OF BEHAVIOR

Taboo words:

as ambivalent stimuli, 316-319

contextually produced, 313

perception of, 310-323
Tachistoscopic recognition thresholds:

and familiarity, 284, 285

and hunger, 283-285

and set, 285

{See also Perception)
Thematic Apperception Test and

human motives, 240-253
Thirst:

local theory of, 63-64

and perceptual responses, 285-289

and water consumption, 71, 72
Thwarting:

conflict-induced, 197

criteria of, 198-200

defined, by repetitive responses, 199
by response failure, 199, 200
by response resumption, 199

as drive source, history of, 194-196

by incompatible responses, 197

methods of, 196-198

as motivational variable, 196

by physical barriers, 196, 197

by stimulus elimination, 197
Time errors in illuminance matching,

285-288
Tobacco deprivation and eyelid condi-
tioning, 229, 230
Transitivity in logic of measurement,
51, 52

Unconditioned stimuli:
as drive sources, 233-235
dual role of, in conditioning, 233,
234

Value:

as drive source, 290, 291
incentive -motivation conception of,

291
response-defined, 290-291
Value orientation and perception,

320-322
Variables, psychological, traditional

names for, 24
Verbal commands as drive sources,
182-186
and achievement motivation, 184,

185
and emotionality, 186
and muscular tension, 185, 186
Vicious-circle phenomenon, 357
Vigilance, perceptual, 303-323
Visual exploration drive, 336-338
Visual stimuli, reinforcing properties
of, 332

Water, consumption of, and depriva-
tion time, 71, 72

Word frequency, familiarity defined
by, 312, 313

Yerkes-Dodson law, 91

Date Due
Due Returned Due

/5-'=j. -f

C.Z

The motivation ot iKhaviof, inain
159,4B878mC2

3 12bE D32fl1 b137

KEEP CARD IN POCKET

ResLib

IT IS IMPORTANT THAT *;
CARD BE KEPT IN POCKET '

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Motivation of behavior"

See other formats